Malnutrition in School-Aged Children and Adolescents in Senegal and Cambodia : Public Health Issues and Interventions Marion Fiorentino

Malnutrition in school-aged children and adolescents in Senegal and Cambodia : public health issues and interventions Marion Fiorentino

To cite this version:

Marion Fiorentino. Malnutrition in school-aged children and adolescents in Senegal and Cambodia : public health issues and interventions. Food and Nutrition. Université Montpellier, 2015. English. NNT : 2015MONTS089. tel-01687697

HAL Id: tel-01687697 https://tel.archives-ouvertes.fr/tel-01687697 Submitted on 18 Jan 2018

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés.

Acknowledgement

First and foremost, I wish to express my sincerest gratitude to my thesis directors Dr Jacques Berger and Dr Frank Wieringa for their guidance, enthusiasm, patience and kindness.

I thank Dr Jacques Berger for believing in me right from the start and giving me the opportunity to go to Dakar, as well as his ongoing encouragement to complete this thesis. I thank Dr Frank Wieringa for his supervision and support during my 3 years in Cambodia. They both gave me the freedom to explore the research field on my own while providing me with constructive criticism. I learned a great deal from their vast knowledge and experience.

I sincerely thank Dr Chounn Chamnan for welcoming me in his department, DFPTQ-Fisheries Administration in Phnom Penh. I would also like to thank Mr Guillaume Bastard, my supervisor at GRET Senegal during the first study in Dakar.

This thesis would not have been possible without the great work achieved on the field by our teams in Senegal and in Cambodia. Special thanks go to my co-workers in Cambodia, Dr Marlène Perignon and Mr Khuov Kuong. I also sincerely thank the other co-authors for the peer reviewed manuscripts, especially Dr Marjoleine Dijkhuizen.

I wish to sincerely thank all the Senegalese and Cambodian children who participated in my research studies, as well as their families. I am grateful to teachers, school directors, and staff from the Ministries of Education in Senegal and in Cambodia for helping us to conduct the research in their schools.

I would like to thank Dr Nanna Roos and Dr Pattanee Winichagoon for agreeing to be part of my thesis committee and for their insightful comments. My thanks also go to the research unit Nutripass, the Ecole Doctorale GAIA and the University of Montpellier for making it possible for me to accomplish this thesis.

Last but not least, special thanks go to Jeff, my parents, my sisters, my family and my dear friends from Phnom Penh, for their interest, patience, love and support.

Table of contents

Acknowledgement...... 3 Table of contents...... 4 Tables and figures ...... 6 Context and objectives of the research ...... 8 Development of the research ...... 11 Chapter 1. Nutrition among school-aged children and adolescents in developing countries: Literature review ...... 13 1.1. Growth, brain growth, sexual maturation and early pregnancy during school age and adolescence leads to specific nutritional requirements: what are the consequences of malnutrition in school-aged children and adolescents? ...... 14 1.2. Prevalence of malnutrition worldwide: is nutrition a public health issue among school- aged children and adolescents? ...... 23 1.3. What are the structural and conjectural determinants of malnutrition in school-aged children and adolescents from the developing world? ...... 25 1.4. How to evaluate malnutrition in school-aged children and adolescents worldwide? ...... 27 1.5. Review of coverage and cost-effectiveness of school feeding programs and staple food fortification: how to improve nutrition among school-aged children and adolescents? ...... 33 Chapter 2. Nutritional status and its dietary determinants of urban African school-aged children and adolescents: Case study in Senegal ...... 43 2.1. Anthropometric and micronutrient status of school-children in an urban West Africa setting: a cross-sectional study in Dakar, Senegal (Published paper) ...... 44 2.2. Nutrient intake is insufficient among African urban school-aged children and adolescents: results from two 24-hours recall in primary state schools in Dakar, Senegal (Manuscript, drafted) ...... 55 Chapter 3. Evaluation of determinants and consequences of malnutrition among school children: data from Cambodia case ...... 73 3.1. Stunting, poor iron status and parasite infection are significant risk factors for lower cognitive performance in Cambodian school children (Published paper) ...... 74 3.2. Height, zinc and soil-transmitted helminthes infections in school children: a study in Cuba and Cambodia (Published paper) ...... 87 Chapter 4. Effectiveness of multi-micronutrient fortified rice through a school feeding program: Case study in Cambodia ...... 95 4.1. Impact of multi-micronutrient fortified rice on hemoglobin, iron and vitamin A status of Cambodian school-aged children: a double-blind randomized controlled trial (Published paper) ...... 96 4.2. Effect of fortified rice on cognitive performance in Cambodian school-aged children depends on premix composition and cognitive function tested (Manuscript, drafted) ...... 111

Chapter 5. Evaluation of currently used indicators for malnutrition using data from Cambodia and Senegal ...... 125 5.1. Current MUAC cut-offs to screen for acute malnutrition need to be adapted to gender and age: the example of Cambodia (Published paper) ...... 126 5.2. Subclinical inflammation increases plasma transferrin receptor and ferritin and decreases plasma RBP but does not affect plasma zinc in school children and women in Cambodia and Senegal (Manuscript, drafted) ...... 136 General conclusion ...... 151 Authors affiliation ...... 152 Synthesis ...... 153 Synthèse ...... 164 References ...... 174 Abstract ...... 197 Résumé ...... 197

Tables and figures Table 1 Public health (Ph) significance of nutritional issues 124-129 ...... 23 Table 2 Evaluation of anthropometric status of Cambodian and Senegalese School-aged children and adolescents ...... 30 Table 3 Indicators of micronutrient status at population levels as defined by WHO ...... 32 Table 4 Effectiveness of rice fortification among school children in developing countries ...... 36 Table 5 Technologies to produce fortified rice kernels 178 ...... 38 Table 6 Recommended nutrients and nutrient levels (mg/100 g of rice) for rice fortification where rice consumption is 150-300 g per capita per day by de Pee ...... 39 Table 7Anthropometric and biochemical status of participants for all and disaggregated for children (<10y) and adolescents (≥10y ) ...... 49 Table 8 Public health significance of nutritional disorders in children from primary state schools of Dakar ...... 51 Table 9 Recommendations of daily intake of macronutrients and micronutrients for populations of children 4-18 y...... 58 Table 10 Mean energy and macronutrient daily intake (adjusted for within-person variability) and prevalence of insufficient and excessive macronutrient daily intake ...... 60 Table 11 Mean micronutrient daily intake (adjusted for within-person variability) and prevalence of insufficient and excessive micronutrient daily intake ...... 61 Table 12 Crude and age-gender adjusted odds ratios (or) and confidence interval (ci) for insufficient nutrient intake associated with micronutrient deficiencies ...... 62 Table 13 Characteristics of school-children participating in the study ...... 79 Table 14 Factors associated with cognitive performance in RCPM test among participating school children ...... 81 Table 15 Factors associated with cognitive performance in Picture completion test among participating school children ...... 82 Table 16 Univariate and multivariate analysis of factors associated with poor performance in Block design test among participating schoolchildren ...... 83 Table 17 Factors associated with cognitive performance in RCPM test before and after adjustment on socio-economic status in a sub-sample of school children (n=616)...... 84 Table 18 Characteristics of the study populations...... 91 Table 19 Zinc and height for age in STH infected and uninfected children...... 92 Table 20 Linear regression models of height for age by STH infection and zinc...... 92 Table 21 Linear regression models of zinc by STH infection...... 93 Table 22 Micronutrient contents of the fortified rices per 100g of uncooked blended rice. URO: UltraRice original formulation, URN: UltraRice new formulation...... 99 Table 23 Baseline characteristics of all children participating in the study and for each intervention group ...... 102 Table 24 Risk factors for anemia at baseline ...... 104 Table 25 Biochemical outcomes and effect sizes after 3 and 6 months of intervention for all participating children ...... 105 Table 26 Biochemical outcomes and effect sizes after 3 and 6 months of intervention for the sub-sample of children with no inflammation at baseline, midline and endline ...... 106 Table 27 Prevalence of marginal VA status after 3 and 6 months of intervention among all children...... 107 Table 28 Micronutrient composition of uncooked rice per 100g of blended rice ...... 114 Table 29 Characteristics of children at baseline (with available cognition data) ...... 118

Table 30 Cognition outcomes and effect sizes after 6 months of intervention for all children .. 119 Table 31 Effects of risk factors for low cognitive scores on the 6 months intervention ...... 120 Table 32 Age and gender characteristics of the participants and proportion of children suffering from acute malnutrition ...... 129 Table 33 Validity of actual WHO cut-off for severe and acute malnutrition in children <5 y ...... 129 Table 34 New cut-offs* by age group and gender for severe and moderate acute malnutrition for children from 0 to 14y ...... 131 Table 35 Demographic and biochemical characteristics of participants: school children from Senegal, school children from Cambodia, and WRA from Cambodia ...... 140 Table 36 Spearman's correlation coefficient (ρ) between inflammatory and micronutrient status variables, in Senegalese school children, Cambodian school children and Cambodian WRA ..... 141 Table 37 CFs and ratios of ferritin by inflammatory status in Senegalese school children, Cambodian school children and Cambodian WRA ...... 142 Table 38 CFs and ratios of transferrin receptor by inflammatory status in Senegalese school children, Cambodian school children and Cambodian WRA ...... 143 Table 39 CFs and ratios of RBP by inflammatory status in Senegalese school children, Cambodian school children and Cambodian WRA ...... 144 Table 40 CFs and ratios of zinc by inflammatory status in Senegalese school children, Cambodian school children and Cambodian WRA ...... 145 Table 41 Effect of correcting TFR, PF, RBP concentrations on the prevalence of low iron status and low vitamin A status, in school children from Cambodia, school children from Senegal and WRA from Cambodia ...... 146

Figure 1 Determinants and consequences of adolescent pregnancies in the developing world: conceptual framework ...... 22 Figure 2 Gender-related differences in prevalence of abnormal status ...... 53 Figure 3 Prevalence of insufficient macronutrient intake according to age and gender group ... 63 Figure 4 Prevalence of insufficient micronutrient intake according to age and gender group .... 64 Figure 5 Unadjusted and adjusted prevalence of insufficient and excessive intake ...... 65 Figure 6 Trial profile. THR: take-home ration, URO: UltraRice original formulation, URN: UltraRice new formulation, WFP-SMP: World Food Programme school-meal program ...... 98 Figure 7 Study design ...... 113 Figure 8 Optimal cut-offs for acute malnutrition (AM) and severe acute malnutrition (SAM) by age group and gender ...... 132

Context and objectives of the research

According to FAO, 870 million people were undernourished in 2010 and 2012. Albeit a global reduction in almost all regions during the previous 2 decades, the reduction rate globally slowed since 2007-2008, consequently to the global financial, economic and food price crisis in 2008, failing to meet the MDG1 by 2015 1. Undernourishment indicates insufficient caloric intake, based on food availability of countries, which does not necessarily reflect malnutrition, while malnutrition is measured by epidemiological studies conducted on representative samples of populations using anthropometric and biological indicators 2. Malnutrition represents a public health issue worldwide. Retardation in linear growth leading to stunting affected 165 million or one third of under-5 children worldwide in 2011, with higher rates found in SEA. In Africa alone 3, 52 million children were wasted, meaning having a body weight too low for their height. Undernutrition among pregnant women increases the risk of intrauterine growth restriction (IUGR) and low birth weight, which are associated to neonatal mortality and stunting. Suboptimum breastfeeding increases the risk for mortality. Stunting is associated with impaired physical and cognitive development of children. Vitamin A, zinc, iodine and iron deficiency account for 12% of preschool children deaths 4.Iodine deficiency is responsible for mental retardation and impaired cognitive function 5. Iron deficiency leads to microcytic anemia, which impairs immune and endocrine function, and decrease work capacity 6. Zinc deficiency contributes to stunting and to increased morbidity 7. Vitamin A deficiency is the leading cause of preventable pediatric blindness and increase morbidity 8,9 . Folic acid deficiency in pregnant women can lead to neural tubes defects 10 . Undernutrition, including intrauterine growth restriction, suboptimum breastfeeding, stunting, wasting, and micronutrient deficiencies is the major contributor to disease burden among preschool years old. It is estimated to be responsible for the death of 3.1 million child each year, corresponding to half of all child death 3,4 mostly in the developing world. Although prevalence of undernutrition was reduced in the previous 2 decades, in SEA and Africa malnutrition is still a major concern, as prevalence of stunting is increasing in Africa, while the highest absolute number of stunted children is found in SEA 4,11 . Thus, undernutrition among children preschool is widely prevalent and its severe consequences on health and development have hindered the economic development and human capital 10 . Maternal nutritional deficiencies contribute to the intergenerational cycle of malnutrition and poverty, and it is widely acknowledged that the critical period to break the cycle of malnutrition is during the first 1000 days of life: from the conception to 2 years. Therefore, nutrition programs target in priority preschool children and pregnant and lactating women, with most efforts and funding given to interventions such as promotion of early and exclusive breastfeeding, micronutrient supplementation of pregnant women, vitamin A supplementation of preschool children, or promotion of dietary diversity and complementary feeding 12 . WHO recommended early child development programs as the most cost-effective interventions in developing countries, regarding child health as well as poverty and inequalities reduction 13 .

However, there is a growing interest about improving nutrition among school-aged-children (SAC) and adolescents. In opposition to under 5 years old children (U5) U5 1, , children aged 5- 9.9 y will be designated here as “ School-aged children ” (SAC). WHO defines adolescent as individuals between 10 and 19 years, with the cut-off between early (EA) and late adolescence

11 « School children » will be used as a generic term to refer to both school-aged children and adolescents studying in primary schools 8

(LA) at 15 years 14 . In 1996, a symposium already questioned “Adolescent nutrition: are we doing enough?”, arguing that world interest in adolescent health regarding pregnancy and HIV had grown in the previous decade, but that nutrition stayed neglected 15 although elevated prevalence of anemia, stunting and thinness were observed among adolescent. In 2006, WHO published a review about adolescent nutrition in South-East-Asia which reported a large percentage of adolescents suffering from nutritional deficiencies and a prevalence of anemia between 12% and 100% across the countries 16 . A review published in 2010 by Best et al 17 summarized the data of nutritional status of children 5-12 y worldwide. Prevalence of severe or moderate anemia, as well as prevalence of iron, iodine, zinc, and vitamin A deficiencies ranged from 20% to 30%, suggesting an important public health issue among SAC, especially from Africa and SEA. The researchers warned about the lack of data. UNICEF published a report in 2011 “Adolescence: an age of opportunity” asking for more efforts to be done in favor of adolescent health, especially in nutrition, and in adolescent girls.

Malnutrition among SAC also inhibits physical and mental development. Short height and low BMI for age at school age are associated with future reduced work capacity and obstetric complications 18,19 . Micronutrient deficiencies and consequent anemia among SAC may impair their growth, vision, concentration, cognitive performance and immune defenses. Recent evidence highlighted the possibility to improve growth, cognition and health outcomes in SAC by improving nutrition 20-24 . It is known that adequate nutrition and health supports development, growth and cognitive achievement among SAC 25 . Early adolescence is characterized by a growth spurt and physical changes associated with sexual maturation 26 . Recent neurological research brought to light the tremendous development of the brain during early adolescence. The neuronal network is drastically reorganized and the number of brain cells can double within one year 26 .Although most part of the brain are developed during fetal life and infancy, the frontal lobe which control reasoning and executive functions achieves a growth spurt during early adolescence. Therefore, nutritional deprivation occurring during adolescence not only perpetuates the stunting process but it can seriously impair cognitive functions and jeopardize schooling achievement. Moreover, lateness of puberty will extend the adolescence, maintaining individuals in a critical situation of high nutrient requirements. The brain and skeletal growth spurt as well as the appearance of sexual characteristics usually occur earlier in girls compared to boys. Nutrition during adolescence should be particularly monitored because of the risk of pregnancy in developing countries where early marriage is common 27 . Adolescent pregnant girls bear the double burden of their own plus their babies’ growth and development. Therefore, even if the main focus remains to improve nutrition of preschool children, investing in late childhood and adolescence could be an effective way to reinforce the global gains achieved in health and nutrition of young children since 1990 3. Children 5-14 y represent 1.1 billion individuals worldwide, of which 90% live in Low or Middle Income Countries (LMIC) 28 *. Schools could be an effective platform to deliver health and nutrition interventions such as school feeding and deworming programs. School feeding programs have been widely implemented with 68 million children receiving daily breakfast, lunch or snacks 29 . But large discrepancies remain between countries or within countries. School feeding programs globally increased school attendance in developing countries, but benefits on nutrition, health and schooling outcomes lack evidence and could be improved using cost-effectiveness tools such as staple food fortification. During the past century, research about indicators to diagnose malnutrition in human populations has been tremendous. One strategy is to evaluate the adequacy of dietary and

9 nutrient intake through food consumption surveys. In order to compare intake with requirements, this implicates to establish nutrient requirements adapted to the population's needs and to collect accurate data about food consumption. This method does not take into account nutrient absorption, retention and utilization. A more direct strategy consists in evaluating the nutritional status by measuring anthropometry and body composition, micronutrient status using biomarkers, most of involving biochemical analysis on blood, urine, hair or sweat samples, or by functional indicators or even clinical evaluation. Ideally, the biomarkers should reflect functional outcomes, and provide information on the level of severity of deficiency or the level of stores. The selection of indicator is crucial and international recommendations are needed for researchers to adequately evaluate nutritional status and dietary patterns between different populations. Beyond the choice of indicator, the establishment of adequate cut-offs to define deficiency is compelling. Indeed, nutritional status does not only depend on food intake but it is influenced by many individual and environmental factors such age, gender, pregnancy, inflammatory status, parasite infestation or genetics. Internationally recommended indicators of nutritional deficiencies and cut-offs adapted to SAC and adolescents are lacking. The growth reference for 5-19 years children of WHO was updated in 2006 but this latest growth reference was questioned as a standard 30 ). Due to poor hygiene and sanitation in the developing world, SAC are particularly at risk for infections. Inflammation raises nutrient requirements, but also affects concentrations of several biomarkers. Ferritin concentrations for example are widely known to be increased by the inflammatory response and some tools were established to adjust prevalence of iron deficiency for inflammation. But one may question the impact of inflammation on transferrin soluble receptor, which is the other indicator of iron status widely used in epidemiological studies, and so far, very little research about this subject is available. Retinol and RBP are influenced by inflammation too. No internationally recommended cut-off to define inflammation is available. Moreover, plasma/serum retinol cut-offs to determine vitamin A deficiency and marginal vitamin A status currently used for SAC and adolescents were determined for young children.

Development of the research

This thesis focuses on the nutrition of school-aged children and adolescents from developing countries. The thesis is based on studies successively conducted in Senegal and in Cambodia, whose the common main objectives were to define their nutritional status and assess the impact of nutritional interventions on micronutrient deficiencies, health and development in school- aged children (SAC) and adolescents. Consequently the thesis will first review in Chapter 1 the epidemiologic situation of malnutrition in the world and the steps to define sustainable intervention strategies adapted to the specific context of countries that are: i) assessment of the nature and the magnitude of nutritional issues in the target populations ii) evaluation of their determinants iii) definition of potential interventions iv) assessment of their efficacy, their effectiveness and their impact. The Chapter 2 will present the first study carried out in Senegal (2 publications). This study was part of a project of two NGOs (Enda-Graf Sahel and GRET) and a French foundation (Danone Communities) aiming at developing a specific food complement for urban Senegalese schooled children. The expertise of IRD was requested for the micronutrient formulation of the food complement. However, the absence of epidemiologic data regarding the target population led us to propose to conduct a cross-sectional study to determine the nutritional status of children from a representative sample of state schools in Dakar and to evaluate their dietary nutrient intake. The results of this study will be presented in chapter 2 (2 publications). The second study was conducted by IRD in school children in Cambodia, in collaboration with an American NGO (PATH) and an international organization (World Food Program) 2. The main objective of this study was to assess the effectiveness of introducing 3 different micronutrient- fortified rices in school breakfasts on improving nutritional and health outcomes. The results of this study are described in chapter 3 (baseline characteristics of the school children, 2 publications) and chapter 4 (impact of the intervention, 2 publications). Data produced by these 2 studies brought us to an overall brainstorming about public health nutrition issues in school-aged children and adolescents, especially indicators currently used to determine them. Consequently, we reanalyzed the combined data from these 2 studies and the results of these analyses are presented in chapter 5 (2 publications).

2 The author declares no conflicts of interest with any of organizations involved in the projects in Cambodia and Senegal 11

Chapter 1. Nutrition among school- aged children and adolescents in developing countries: Literature review

1.1. Growth, brain growth, sexual maturation and early pregnancy during school age and adolescence leads to specific nutritional requirements: what are the consequences of malnutrition in school-aged children and adolescents?

The growth period arising in adolescence is the second most important in life after the one occurring during the first year of life. This growth spurt dramatically increases nutrient requirements. Adequate intake of energy, protein, calcium, iron, zinc and folate is therefore crucial, especially during the peak of velocity of growth when nutritional requirements may be up to twice those of the remaining period of adolescence 31,32 . Due to different body composition and biological changes, and to earlier puberty timing in girls, nutrient requirements start to diverge between boys and girls at puberty. At menarche, iron needs are tremendously increased in girls due to blood losses 33 . School age and adolescence is also a critical period for neurological development. Even though neurulation, the formation of the receptive language, seeing and hearing brain areas occur during fetal life and before the age of 5 years, synaptogenesis and neurogenesis continues up to the end of adolescence 34 . Peaks of brain development, especially frontal lobes have been found in the first 2 years, 7 to 9 years and in the mid-teenage years 35,36 . Therefore, the development of the full range of executive functions is suggested to occur in late childhood and adolescence as myelination of the frontal lobes proceeds 37 . Peaks of brain development occur at different timing according to the brain area : frontal and parietal gray matter growth peaks at 12 for boys and 10 for girls, while white matter volume increase linearly, and peaks of temporal gray matter growing occur at the end of adolescence 38 .

1.1.1. Consequences of acute and chronic malnutrition during school age and adolescence

Stunting was found to predispose to chronic metabolic diseases obesity or overweigh in late childhood or in adulthood, probably because the rate of lipid oxidation is lower in stunted children, leading to more central fat accumulation 39 . Stunting is also associated with long-term adverse effects such as impaired cognitive performance. Stunting is mainly the result of chronic nutritional deprivation that occurs before the age of 3 years, which is hard to recover later in life. However, some research suggests that health and nutrition interventions can slow down or even reverse the stunting process in SAC 21,24,40 . Stunting had adverse consequences at adult age. Low women height is associated with small pelvic size, which increases risk of obstetric complications during delivery. Reduced height was also found to impair work capacity 18 . Thinness reflects recent nutritional deprivation 17 . Thinness during school-age and adolescence may delay pubertal maturation. Thinness is a particular concern among adolescent girls which are more likely to give low birth weight babies. Muscular strength and work capacity is more likely to be reduced in individuals who were thin during school age 18 . Obese or overweight SAC are more at risk for type 2 diabetes, high blood pressure, metabolic syndrome and psychological disorders 41 . Dyslipidemia and hyperinsulinemia associated with the metabolic syndrome track throughout the lifetime and are associated with cardiovascular disease 42,43 .

1.1.2. Consequences of micronutrient deficiencies during school age and adolescence

1.1.2.1. Iron role and deficiency

The main role of iron is oxygen transportation, as a component of hemoglobin. Worldwide, iron deficiency is the main cause of anemia, which symptoms as fatigue, breathlessness and headaches that can impair physical activity and concentration among SAC 44,45 . Moreover, iron deficiency and iron deficiency anemia was showed to increase the risk of infection as well as cognitive impairment 45 . Therefore the implication of iron deficiency in anemia leads to increased morbidity, absenteeism, and indirectly to impaired school achievement. Including SAC in epidemiological studies concerning anemia has been recommended 46 , but is often not done. One major concern about iron deficiency among SAC and adolescents is its adverse effects on cognitive functions. Iron deficiency and iron deficiency anemia are consistently associated with impaired cognitive function and lower school performance in SAC 47-51 . A study among 12000 Taiwanese SAC and adolescents showed increased risk for mood disorders, autism spectrum disorder, attention deficit hyperactivity disorder, and developmental disorders associated with iron deficiency anemia 52 . The lower scores in cognitive functions with ID or IDA could be the result of hematologic impairment during early childhood, which adverse effects on cognition perpetuates at school age. However, studies reported improved cognitive scores or educational achievement or improved motor development in response to iron supplementation among SAC with ID, IDA or normal iron status 22,45,53-55 . As an example, a meta-analysis of randomized controlled trials of daily iron supplementation in children 5-12 years reported improvement of height, global cognitive scores, measures of attention and concentration. Among anemic children, daily iron supplementation improved intelligence quotient and weight 56 . This indicates that some of the cognitive deficit resulting from iron deficiency at school age can be reversed, In contrast to infants, where long-terms effects of iron deficiency may be permanent 57,58 . The adverse effects on cognitive and educational performance of IDA in school-age children appear more transitory than in infants, where cognitive impairment is less reversible 45 . Iron supplementation was also found to improve the beneficial effect of deworming on school attendance 59 . The brain is sensitive to iron deficiency because iron plays a role in neurotransmission, especially in the dopamine pathway, and in homeostatis regulatory mechanisms 36,60 . Iron deficiency alters myelination, neurotransmitter synthesis and hippocamp energy metabolism 61 . Therefore, iron is a key nutrient in the development of executive functions 36 .

1.1.2.2. Iodine role and deficiency

Iodine is crucial for cerebral growth and development because it is required for the production of the thyroid hormones triiodothyronine (T3) and thyroxine (T4). Long-term and severe iodine deficiency causes cretinism and goiter. However, even at less severe stage, iodine deficiency can be responsible for lower cognitive performance, deaf-mutism, or birth defects. Iodine deficiency among pregnant women and infants is the 1 st cause of preventable mental retardation in children 44 . A meta-analysis reported that SAC living in iodine-deficient areas have intelligence quotient 13 points lower than children from iodine-sufficient regions ( 62,63 . Some studies report positive correlation between cognitive performance and urinary iodine 64,65 , with also evidence of iodine supplementation improving cognitive functions in SAC 23,64,66-68 . Iodine deficiency has more serious consequences at fetal age, but cognitive functions can be

15 affected across all ages as hypothyroidism as an impact on neuronal development, structure and function 36 . Therefore, some of the cognitive impairment resulting from iodine deficiency still can be addressed at school age.

1.1.2.3. Vitamin A role and deficiency

Vitamin A is a component of the retina of the eye, and plays a crucial role in vision. Severe vitamin A deficiency leads to corneal destruction and blindness, but this condition called xerophtalmia usually peaks at 2 to 3 years of age 17,44 . However, chronic vitamin A deprivation persisting at school age may increase the prevalence of xerophtalmia beyond 5 years 69 . Vitamin A is also involved in the immune response. Because vitamin A deficiency is associated with increased morbidity and mortality in infants and young children, efforts to control vitamin A deficiency are concentrated on pregnant women and young children, thus vitamin A supplementation is not usually recommended in SAC 17,44 . However, studies in Thailand and Colombia revealed that vitamin A deficiency, even mild, increase the risk of respiratory diseases and of diarrhea in SAC 70,71 . Some research suggest a role of vitamin A in growth during later childhood and in sexual maturation, making adolescent particularly vulnerable to vitamin A deficiency 72,73 . Plus, vitamin A deficiency can have adverse effects on iron status, with vitamin A supplementation contributing in the prevention of anemia 73 . A study on anemic Tanzanian SAC showed improvement of growth and hemoglobin status with vitamin A supplementation 74 . Therefore, vitamin A status should be monitored among SAC and adolescent, because of the potential adverse consequences on health and schooling outcomes linked to the risk for eye disorders, increased morbidity and ceased growth. But overall, data about the impact of vitamin A deficiency or the benefits of vitamin A interventions on health outcomes such as growth velocity and eye functioning among SAC and adolescents is lacking.

1.1.2.4. Zinc role and deficiency

Zinc plays a role in multiple aspects of metabolism, and zinc deficiency results in reduced growth rather than in specific clinical signs 75 . Therefore, due to the growth spurt during adolescence, zinc status is particularly worrisome among this age range. Indeed, zinc supplementation improved growth velocity among stunted children and adolescents in 3 studies 76-78 . Zinc deficiency is also associated with impaired immune defenses and increased prevalence of diarrhea 44 . The beneficial effect of zinc supplementation to prevent acute diarrhea and pneumonia and to reduce mortality was demonstrated among under 5 years old children, but such evidence among SAC or adolescent is scarce 79-81 . However, zinc supplementation improved cognitive performance and taste acuity among adolescent Indian girls 82 .

1.1.2.5. Other micronutrients role and deficiencies

Iron, iodine, zinc and vitamin A are the most studied micronutrients 17 as deficiencies are very prevalent around the world, and their adverse effects on health and development on children are well described. However, in the present review of literature, I will also focus on less considered vitamins and minerals such as thiamine, vitamin E and copper.

Thiamin (Vitamin B 1) is found in vegetal and animal sources, mainly in the form and outer coats of unrefined cereals, but also in green vegetables, fish, meat, fruit and milk. It is highly water soluble thus rice should not be washed 44 . Thiamin diphosphate is cofactor for some enzymes involved in glucose and carbohydrates metabolism. Thiamin triphosphate has functional properties in the neuronal membrane. Thus, thiamin has a crucial role in the development and the maintenance of neuronal functions 83 especially in frontal loves and basal ganglia 84 . Thiamin deficiency results in neuronal losses related to brain energy deficit due to lower activity of thiamine dependent enzymes and oxidative stress affecting carbohydrate metabolism. The most known consequence of thiamin deficiency is the neurological and cardiovascular disease beriberi. The dry beriberi is associated with motor and sensory disorders while the wet beriberi symptoms include edema, mental confusion and hear disorders. Infantile beriberi occurs in children breastfed by thiamin deficient mothers and cause cardiac disorders, aphonia and sudden death ( 44,85 . Other signs of thiamin deficiency are weakness, fatigue, psychosis 44 . Thiamin deficiency during infancy results in long lasting language disorders 86 . Low thiamin intake in adolescent results in increased risk of aggressiveness and delinquency 87 . Some research suggest a possible role of thiamin deficiency in the apparition of tumors 88 . A study reported that low intake of vitamin B1, B2, B3, B5, B6 is associated with aggressiveness and delinquency while low intake of vitamin B6 and folate is associated with withdrawnness and depression.

Copper is mainly found in seafood, organ meats, nuts, beans, dried fruits and vegetables, black pepper, cocoa, and dark green leafy vegetables. Intestinal absorption (30-40%) is regulated according to the intake. Clinical features of copper are mostly related to cuproenzymes activities (metalloproteins). Copper is involved in metabolism, angiogenesis, oxygen transport, and antioxidant protection, including copper zinc superoxide dismutase 89 . Copper deficiency main clinical manifestations are hematologic, with lower white blood cells, which reduces immune response, and neutropenia. Mechanism of reduction of neutrophils is not clear, but hypotheses include destruction of cells in the bone marrow or in the circulation, impaired cell differentiation, proliferation, or increased cellular destruction. Both high and low serum copper levels were found to be associated with unexplained anemia 90 . Women have generally intake below their requirements, which may have harmful effect during pregnancy such as spontaneous abortion, bone, immune and neurological abnormalities or increased risk for cardiovascular disease in later life 89,91 . Nervous damages, hardly distinguishable from vitamin B12 neurological manifestations, such as neuropathies, or myelopathies can occur at severe stage of copper deficiency 92 . Copper deficiency has 2 adverse effects on iron metabolism. When iron and copper deficiencies are concomitant, it causes hypochromia which continues even after treatment with copper alone. As a later effect, even after iron stores are repleted, erythropoiesis will stay impaired 93 . Increased severe respiratory infection prevalence was observed in copper-

17 deficient infants. Following hematological manifestations as anemia and neutropenia, copper deficiency can cause osteoporosis, damage nerve cords and result in spasticity.

Vitamin E is exclusively provided by the diet, and is the major lipid-soluble antioxidant in the cell antioxidant protection system. Its most interesting form from a nutritional point of view is alpha-tocopherol. Vitamin E is essential to protect cell membranes and low-density lipoprotein from oxidation by free radicals 85 . But some research suggests a wider biological role for alpha- tocopherol, including regulation of gene expression and coordinated movement through cerebellar cells. However more research is needed. Early signs of vitamin E deficiency include leakage of muscle enzymes and increase erythrocyte hemolysis 85 which can cause anemia 94 In human and animal models, vitamin E deficiency was shown to induce myopathies, neuropathies and retinal damage 85,94,95 . Neuropathy and anemia may be caused by excessive free radical damage to the axons of sensory neurons and to the red blood cell membrane 94 . Prolonged vitamin E deficiency during infancy impair cognitive functions 96 . Some research highlights the protective role of vitamin E against heart disease but evidence remains inconsistent 85,97 . Some research showed that in adolescents and in children, poor vitamin E? status is associated with stunting 94,98 .

Effects of micronutrient deficiencies on cognitive functions are described in sections 3.1 and 4.2.

1.1.3. Focus on adolescent girls, a vulnerable population due to the risk of pregnancy

1.1.3.1. Magnitude of childbearing by adolescent girls in the developing world

The number of adolescents (10-19y) living in developing countries is estimated to be 1.1 billion. They represent 88% of the total population of adolescents and 19% of the population of developing countries. Half of them live in South Asia, East Asia of pacific region 26 . Although marriage during adolescence had declined in the past decades, especially in Latin America and in North Africa 99 still 21% of girls aged 15-19y are currently married in the developing world, with the rate reaching 30% in the least developed countries 26 . The highest rates of marriage of girls before 15y are found in South East Asia, India and sub-Saharan Africa, with respectively 53%, 38% and 35% of girls being married before 15 y in Bangladesh, in Niger and in Chad respectively 27 . As marriage is generally followed by childbearing, 20% of young women gave birth before the age of 18 in developing countries 26 this proportion reaching 66% in Sub- Saharan Africa 100 . With 55 births per 1000 girls 15-19 y, the rate of birth among adolescents is twice higher in the developing countries compared to industrialized countries 26 . 16 million girls 15-19 y and 1 million girls <15y give birth every year, 95% of these births occurring in low and middle income countries 101 .

1.1.3.2. Multifactorial early marriage and pregnancy

Poverty and cultural traditions are powerful factors driving early marriage (figure 1). Marrying off a girl results in one less mouth to feed and ensure that she will be supported. In some cultures, families receive a “bride price” in the form of cattles or money 27 . In insecure environments, marriage of young girls could be seen as a strategy to protect them from rapes

18 and dishonor. In areas where girls education is not valued, marriage and procreation is often the only option for girls 27 . Uneducated girls are more likely to have few alternatives for their future, few skills and self-confidence to become economically dependent. But premature marriage is not only practiced in underprivileged families. The high social value of virginity conduct families to marry their daughters immediately after puberty or even before, in order to protect them from promiscuity and the risk of pregnancy outside of marriage. The social pressure to become pregnant within 1 year after the wedding is high in the developing countries, even if their own physical growth is not terminated 27 . For example, in Cambodia, where 10% of 15-19y girls are married, 8% of them already experienced a pregnancy, and only 15% of married adolescent girls use a modern birth-control method 102 .

1.1.3.3. Health consequences of early pregnancies

Pregnancy in adolescents can have adverse consequences on the fetus and the infant. In low and middle income countries, infants born to mothers under 20 y are at 1.5 - 2 times higher risk to be stillborn or to die before the age of 5 years 101,102 . Reasons for this are numerous. First of all, adolescents in developing countries are likely to enter pregnancy with poor nutritional status. Rates of anemia among girls 15-19y are above 40% in India and some sub- Saharan countries, which indicates a severe national public health issue. Prevalence of underweight in girls 15-19y ranged from 25 to 47% in Sub-Saharan Africa, India and Bangladesh 26 . Nutritional deficiencies in pregnant or lactating women are known to be risk factors with poor pregnancy outcomes and poor nutritional infant status. Maternal iron deficiency and iron deficiency anemia for example are associated with low birth weight 103 . Vitamin A deficiency in lactating women is associated with low vitamin A breastmilk concentration and hence to vitamin A deficiency in their infants 104 . Poor micronutrient status in pregnant women was also found to be related to altered metabolism, organ growth and function, and to elevated risk of chronic disease in their children later in life 105 . Secondly, besides pre-pregnancy nutritional status, pregnancy during adolescence is associated with poor pregnancy outcomes compared to pregnancy in adults even in developed countries. A study in Gabon reported odds ratio >2 for the risk for low birth weight and less frequent antenatal care visits in adolescent pregnant girls compared to adult counterparts but no information about their nutritional status was available 106 . Animal studies reported that initial low maternal BMI as well both overnourishment and undernourishment during pregnancy in adolescents resulted in lower birth weight and placental weight 107 . Because of irregular menstruations and the absence of planning of the pregnancy common among adolescents, adolescent pregnancies are often undetected for late into pregnancy, preventing early start of folic acid supplementation, recommended prior to conception and until the 12 th week of pregnancy. Folate deficiency is a particular concern in pregnancies because deficiency increases the risk for neural tubes defects in the fetus. 108 It was also associated with a higher risk for Small-Gestational-Age (SGA) delivery in pregnant UK adolescents. 109 Another concern is calcium status in adolescent pregnancy. The growth spurt during adolescence may deplete calcium stores. The skeleton is still increasing its density during adolescence and demineralization of bone during this period could be harmful 108,110 . Suboptimal calcium status in pregnant adolescents can both impair maternal as fetal bone growth 110 .

Pregnancy during adolescence has not only harmful effects on infants but also on mothers. Adolescent girls contribute to 11% of births worldwide, but because their bodies are not ready for childbearing, they account for 23% of the global burden of disease related to pregnancy and childbirth 102 . Indeed, girls aged 10-14 years have 5 time higher risk to die from pregnancy and childbirth complications than adult women 102 . Complications related to pregnancy or childbirth such as hemorrhage, sepsis, preeclampsia, eclampsia, are the leading cause of mortality among 15-19 y girls 27,100 . Among the 16 million of adolescent girls giving birth every year, 50,000 die 100 , or 31.3/10,000 deliveries. Maternal-fetal competition for nutrients was first observed in adult women. While well- nourished mothers provide unlimited access to nutrients for their fetuses, undernourished women tend to retain their own tissue nutrient stores at the expense of the fetal growth 111 . In adolescent girls, pregnancy superimposes nutrient requirements for fetal growth on those for maternal growth. Therefore, there is a competition between the mother and the fetus to meet simultaneously and adequately the nutrient requirements for the growth of both mother and fetus. Indeed, even if it was generally admitted that the growth spurt occur before menarche 112,113 , it was showed that growth can continue during years following menarche, especially in undernourished populations were adolescence can be extended, offering an opportunity to catch up for stunting 114-116 . The nutrition partitioning between mother and fetus is modified according to the nutritional status of the mother 108 108 . Some researchers suggest that in case of severe deficiencies, maternal nutrition is favored, possibly through immature placental development, 108 while in moderately malnourished or normally nourished mothers, the fetal compartment is favored 108,117 . A prospective study in rural Bangladesh revealed that pregnancy and lactation among malnourished adolescents resulted in weight loss and depletion of fat and lean body mass. The linear growth was hampered between 0.6 to 2.7 cm in pregnant girls compared to non-pregnant girls. This phenomenon is suggested to only occur in malnourished girls. The negative effect of pregnancy on anthropometric outcomes was larger among adolescents who became pregnant less than 2y after their menarche. 2 hypotheses were suggested. Impairment of growth and mass gain may be related to the competition for nutrients between the teenage mothers and the fetus related to insufficient nutrient intake. Or among adolescent, further elevation of oestrogen level during pregnancy may accelerate the epiphyseal closure, therefore terminating bone growth 118 .

Other research reported that nutrient partitioning is regulated by maternal, placental and fetal endocrine hormones, and is altered in pregnant adolescents where maternal growth is promoted at the expense of the increasing nutrient requirements of fetal and mammary gland development 119 . In normal pregnancies, fat stores usually increase during the 1st two trimesters, but decrease during the third trimester and immediate postpartum 120 . One study reported increasing maternal fat stores late in pregnancy among adolescent mothers still experiencing growth, which was related to lower birth weights. Even though they were accumulating more fat in late pregnancy and had larger gestational gains compared to non- growing pregnant adolescent or to pregnant adults, they gave births to lower birth weight infants. Moreover, birth weight was correlated to maternal energy intake only in still-growing mothers, suggesting a higher reliance of the fetus on the maternal diet than on maternal fat lipolysis for energy supply. Although weight gain and accumulation of nutrient stores seemed sufficient, mobilization of stores in late pregnancy was suggested to be reserved to maternal

20 growth instead of fetal growth. No data about their nutritional status were available but around 20% of them had insufficient energy intake 121 . Hyperinsulinemia is known to restrict placental blood flow and to diminish lipolysis by reducing blood flow to adipose tissue. Researchers suggested that nutrient competition among pregnant adolescents may be regulated through hyperinsulinemia and insulin resistance to reduce maternal tissue uptake of glucose and to maintain adequate maternal glucose concentrations 121 . Also, oxidation of maternal glucose which provides fetal energy may be different if the mother is adolescent or adult. Adolescents may increase glucose production primarily through glycogenolysis while their adult counterparts do so through gluconeogenesis. Plus, it appears that pregnant adolescents do not increase glucose production by the same magnitude as pregnant adults 122 . Over- nourishment of pregnant adolescents may also adversely impact pregnancy outcomes. Animal studies reported a significant reduction of placental mass, birth weight, duration of gestation, quality and quantity of colostrum, and increased occurrence of spontaneous abortion in over nourished pregnant adolescents compared to normally nourished counterparts. This negative effect of overnourishment is probably caused by reduced placental growth 119 . Impairment of pregnancy outcomes in overnourished adolescents is related in early pregnancy to impaired proliferation of fetal cells, uteroplacental blood flows and angiogenesis. In later stages of pregnancy, placental mass is reduced in overnourished adolescents compared to normally nourished adolescents, resulting in reduced fetal nutrient uptakes. Similar animal models suggest that in pregnant undernourished adolescent, fetal growth is also reduced despite normal placenta mass, while maternal stores and transplacental glucose gradient are diminished 123 . Nutrient partitioning orientation in pregnant adolescents depends on maternal nutritional status and evidence remains unclear. However pregnant adolescents in developing countries have to support the triple burden of delayed growth, poor nutrient intake and childbearing, which can have adverse effect on the mother, the child or both.

1.1.3.4. Social Consequences of early pregnancies

Early marriage and childbearing may also have dramatic social consequences. Indeed, married adolescents usually drop out school and do not go back to school as they are expected to be at home to take care of the house, husband and children. Because of their poor education and the sexual past of their husbands they are also more exposed to HIV and sexually transmitted disease compared to their non-married counterparts which are likely to no have started their sex life. Because of their inexperience and their poor ability to manage a home and a family, young brides and mothers are also more likely to be divorced and abandoned, increasing their financial and social insecurity 27 . Premature marriage and childbearing therefore dramatically contributes to the intergenerational cycle of malnutrition and poverty. Multifactorial health and social causes and consequences of premature pregnancies in developing countries are summarized on figure 1.

FIGURE 1 DETERMINANTS AND CONSEQUENCES OF ADOLESCENT PREGNANCIES IN THE DEVELOPING WORLD : CONCEPTUAL FRAMEWORK

1.1.3.5. Interventions to prevent early marriage and early childbearing

Consequently, specific nutrient intakes requirements for pregnant and married adolescent girls, taking needs for both adolescence and pregnancy into account, should be established 100 .Nutritional interventions for pregnancy and pre-pregnancy such as micronutrient supplementation should focus in priority on married adolescents. Social interventions to delay age at marriage and to promote family planning should be implemented among adolescents.

1.2. Prevalence of malnutrition worldwide: is nutrition a public health issue among school-aged children and adolescents?

1.2.1 Chronic and acute malnutrition

A review about literature from 2002 to 2009 reported prevalence of malnutrition among children aged 6 to 12 years from Latin America, Africa and Asia. In average, prevalence of stunting was 22% in Africa, 29% in SEA, and 16% in Latin America. Stunting was above 20%, indicating a public health issue, in rural school-aged children in India, Nepal, and Laos. High prevalence of stunting, between 30% and 74% were reported at national level in Guatemala, North Korea, Madagascar, Malawi and Vietnam. The review indicates average prevalence of thinness of 35% in Africa and SEA, and 6% in Latin America. According to national surveys, between 35% and 50% of SAC were classified as thin in Sri Lanka, Vietnam, Madagascar and Uganda. Severe prevalence of thinness, ranging from 77% to 90% was observed in disadvantaged settings of eastern India, Bangladesh and rural South Africa.

TABLE 1 PUBLIC HEALTH (P H) SIGNIFICANCE OF NUTRITIONAL ISSUES 124-129

Nutrition Indicator Prevalence Public health significance issue low height for Stunting ≥20% PH issue age z-scores low 10-20% moderate PH issue serum/plasma retinol level ≥20% severe PH issue Vitamin A ≥5% mild PH issue deficiency night blindness 1-4.9 % moderate PHissue 0.01-0.99% severe public health problem 5-20% mild PH issue Anemia low Hb level 20-40% moderate PH issue ≥40% Severe PH issue Vitamin B12 low serum ≥5% public health issue deficiency vB12 low ferritin <20% and high TFR < 10% ID is not prevalent ID and inflammation are low ferritin < 20% and high TFR ≥ 10% Iron low ferritin and prevalent deficiency high TFR low ferritin ≥ 20% and high TFR < 10% ID is prevalent low ferritin ≥ 20% and high TFR ≥ 10% ID is prevalent

1.2.2 Iron, vitamin A, zinc, and iodine deficiencies

Only the prevalence of vitamin A, zinc, iodine and iron deficiency and anemia were reported in the review by Best et al due to lack of studies available reporting other micronutrient deficiencies. Most studies on malnutrition of SAC and adolescents reports anemia and iodine deficiency 17 . Prevalence of iron deficiency and iron deficiency anemia was in average respectively 29% and 19% in Africa, 20% and 4% in SEA, 14% and 9% in Latin America. Iron deficiency was very high in Cote d’Ivoire , with 59% of rural SAC being iron deficient. More data are available about anemia, which was in average 29% in Africa, 32% in SEA and 14% in Latin America. In Africa, prevalence of anemia was below 10% in Rwanda and Ethiopia at national levels. But in so me regions of Cote d’Ivoire, Kenya and Mali, prevalence of anemia was above 40%, indicating a severe public health issue, which was similar in rural areas of India, Philippines, Vietnam and Cambodia and in some urban and rural areas of India. National surveys reported prevalence between 23% and 38% in Thailand, the Philippines, Nicaragua and Colombia 17 . However, in some Asian settings, hemoglobinopathies was a greater contributor to anemia than iron deficiency 46 . Thus, the etiology of anemia should be carefully examined before designing any iron-containing intervention to reduce the prevalence of anemia.

The review does not report median urinary iodine concentrations (UI), which is used as criteria to assess the magnitude of IDD in population of SAC. However, prevalence of iodine deficiency defined as children below 100 µg/L was on average 33% in SEA and in Africa, and 14% in Latin America. Prevalence between 60% and 90% were reported in several African and Asian countries at national level. National surveys in Latin America reported prevalence below 10% in most cases, but in some areas of Guatemala, Bolivia and Mexico, prevalence ranged from 40% to 80%. These findings indicate that iodine deficiency is prevalent in SAC from developing countries, and is not necessarily limited to children from poor settings 17 . Prevalence of vitamin A deficiency among SAC was in average 32% in Africa, 17% in SEA, and 9% in Latin America. Prevalence of vitamin A deficiency ranging from 50% to 90% indicating severe public health issue was observed at national level in the Philippines, and in subpopulations of SAC of Bostwana, South Africa, Kenya, Burkina Faso and India. Data about zinc deficiency are scarce. The review of Best et al suggested an average prevalence of zinc deficiency around 54% in Africa, 49% in Latin America, and 29% in SEA. Zinc deficiency ranged from 19% to 57% in most studies conducted in Africa, Latin America and SEA 17 .

1.2.3 Other micronutrient deficiencies

Beriberi is known to be still prevalent in rural Asia because of the habit of polishing of rice, removing the outer core which contains vitamin B1. It affects principally infants and alcoholic individuals. However, prevalence of thiamine deficiency of 90% was reported in Indian SAC 130 . A national survey in Taiwan reported around 10% of marginal thiamin status and 8% of thiamin deficiency in SAC 131 . Hookworm infection, which is common in developing countries 132 , is a risk factor for thiamine deficiency 85 . Figures about thiamine status in SAC and adolescents from developing countries are very scarce. Thiamin deficiency was common among sick infants in Lao 133 . Thiamine fortification of staple foods such as flour was shown to be an effective simple and safe strategy to improve thiamine status 134 .

1.3. What are the structural and conjectural determinants of malnutrition in school-aged children and adolescents from the developing world?

1.3.1 Impoverished children living in urban areas are also vulnerable to nutritional deficiencies

It is widely recognized that undernourishment is more prevalent in rural areas compared to urban areas. A review of evidence from 36 developing countries in 2005 reported consistently lower stunting and wasting prevalence in urban compared to rural areas, with HAZ being 0.5 higher, and prevalence of stunting significantly lower in most cases. Socioeconomic determinants of nutritional status were similar in urban and rural areas. In average, children in urban areas benefit from higher women education, better availability of water and sanitary facilities and higher socioeconomic status which all have a positive impact on their growth, and to a lesser extent on their weight gain. However, the authors also highlighted the enormous gap within urban areas, arguing that children from poor urban areas may be just as vulnerable as rural children 135 . Furthermore, a study in 2005 among 15 sub-Saharan countries reported that the discrepancy between rural and urban areas has tended narrow in the past 2 decades, due to an increase in the prevalence of urban malnutrition, and that this urban-rural difference became negligible after adjusting for SES status 136 . Indeed, even if health and sanitation service are more available in urban areas, differences between socioeconomic groups within urban areas are tremendous, with the poorest households living in slums at high health risks. This may explain, at least partly, similar stunting rates between the poorest in urban areas and the poorest in rural areas 137 . Living in urban areas is characterized by a greater reliance on cash income for food and non-food purchase, making urban residents more dependent on employment and labor. Poor urban households become heavily vulnerable to price and income changes. In Africa, urban populations tends to consume more rice and wheat, which are internationally traded, at the expenses of traditional cereals and roots 138 . This makes poor urban residents particularly exposed to the international market variations. Breastfeeding was reported to be 4-6 months shorter in urban areas than in rural areas 138 and overall, research suggests that mothers in poor urban settings combine their income-generating activities with their child care responsibilities. Their efficiency depends on their options for alternative childcare. The limited family network in urban areas may make women more isolated. The poorest will then take their child to their places of work, which can have adverse consequences on their income 69 . Consequently to urban lifestyle and increased female employment, home- prepared food consumption decrease at the expense of street food or processed food which are usually of poor nutritional and sanitary quality 139 . 138 . One of the coping strategies of poor urban households is urban agriculture. A positive association was found between engagement in urban agriculture and dietary adequacy indicators 140 as it can be an importance source of micronutrient rich foods for the family if it is consumed rather than sold 138 . However, urban agriculture is currently a negligible part of agricultural production and income, although it could play a substantial role in urban poverty and food insecurity reduction 141 .

1.3.2 The food price crisis and the global economic crisis intensified food insecurity and malnutrition among under-privileged populations, especially in urban areas

The world witnessed a fuel and food price crisis in the 2000’s, followed by a global financial and economic crisis in 2008, which had a worldwide impact on food security and nutrition. In 2008, the price of the food basket was 48% higher (20% in West and Central Africa, 30% in Latin America, 45% in Asia, 68% in East and East and Southern Africa) compared to the 5 previous years 142 . Despite the decrease of food prices in 2009 following the global financial crisis, all developing countries suffered from a decline in energy consumption between 2005 and 2010, resulting in potential 450,000 more people affecting by hunger 142 .These numbers do not account for coping strategies 142 . It is known that vulnerable households first reduce the diversity and the quality of food before diminishing their energy intake 143,144 . For example, in the context of the food price crisis of 2008, street food was increasingly preferred because of their cheap price resulting from economies of scale 145 . Therefore, the price food crisis may not have resulted in lower energy intake but in lower micronutrient intake. In both cases, it is likely that the food prices crisis had increased malnutrition rates in the developing world, especially among urban poorest households 142 . A study in Bangladesh reported an increase in the prevalence of wasting in 24-59 months children of 5.5% in rural areas, and of 6.7% in urban areas, with a stronger effect among the poorest households 146 . Another study suggested that economic crises impaired the vitamin A status of mothers and young children in vulnerable populations, and that, the prevalence of vitamin A deficiency and night blindness may not have declined even after food prices declined after 2008 147 . The recent food price crisis pushed many poor households to put unemployed mothers or children on the job market, which impaired child care and school attendance 148 . Consequently, the food price crisis has increased the vulnerability of poor urban households to food insecurity, while the global crisis caused unemployment and decreased incomes, leading to global increase of poverty and malnutrition. The nutritional status of SAC and adolescents living in poor urban settings may have been affected also but data is scare.

1.3.3 The transition between preschool age and school age

The effect on nutritional status of the transition from preschool age to school age is not well described. One can assume that under 5 years old children spend more time under the supervision of their mother or caretaker. Under 5 years old children are more likely to benefit from maternal feeding and health care than SAC. Indeed in Senegal, at the time children are enrolled to school, mothers often give a small amount of money, for their breakfast or snacks 145 . SAC consumed less food prepared at home than their younger counterparts. This could explain a degradation of their nutritional status at school age, which is aggravated with age, as their autonomy grows. This phenomenon could be characterized by a low prevalence of chronic malnutrition and a high prevalence of acute malnutrition. Indeed, street food or processed food that children and adolescents buy with their small allowance money often present poor sanitary and nutritional value 149,150 .

1.3.4 Double burden of malnutrition among school-aged children and adolescents from developing countries

The prevalence of overweight was on average lower in SEA and Africa compared to Latin America with prevalence of overweight including obesity being respectively 7%, 13% and 26%. Most studies in Latin America were conducted in urban settings, but at the national level, the prevalence of overweight was between 20% and 30% in Mexico and in Brazil, and the prevalence of obesity was 9% in Mexico, and 21% in Chile, suggesting that overnutrition is not restricted to urban populations in Latin America. In urban China and India, overweight was found around 25% and obesity between 12% and 21% 17 . In the past decades, many developing countries have experienced a nutrition transition related to massive urbanization and globalization and characterized with increasing consumption of energy, saturated fats, refined sugars and salt, increasing the risk of obesity and associated chronic diseases 138 . In 2006, Popkin reported that dietary changes, activity patterns, and large- scale decreases in food prices have contributed to the emergence of non-communicable diseases in urban areas of developing countries. New diets are characterized by higher intakes of animal protein and partially hydrogenated fats and lower intakes of fiber 141 . As seen above, the food price crisis and the global crisis may have blocked this tendency of increasing overnourishment. Overweight is much more prevalent in Latin America compared to SEA and Africa. Furthermore, it may be a concern only among higher states of urban societies 151 . A study in China reported that high SES and urban location are positively associated with frequency of intake of high- energy foods 152 . To conclude, malnutrition, encompassing both under- and over-nutrition, can have serious adverse effects on the physical and development of SAC and adolescents. The prevalence of stunting and acute malnutrition, as well as iron, iodine, zinc and vitamin A deficiencies are high among SAC and adolescents in the developing world, especially in Africa and SEA. Structural and conjectural factors such as the transition between the preschool age to the school age, the urbanization, and the food prices and economic crisis in the past decade may explain poor food habits and nutritional status among SAC and adolescents in the developing world. Four publications in this thesis will illustrate these findings. Results of a cross-sectional survey among children from primary state schools of Dakar are presented in 2 papers, one about the anthropometric and micronutrient status, and the second about nutrient intake derived from 24-hours recall. In addition, 2 papers will describe the determinants of malnutrition in Cambodian school children.

1.4. How to evaluate malnutrition in school-aged children and adolescents worldwide?

1.4.1 Anthropometry

Child growth standards are needed to rank children relatively to others of the same age and sex and to evaluate their growth status. A growth standard represents a recommended pattern of growth associated with specific health outcomes. It should be designed to describe how children should grow in optimal conditions and should not reflect the growth pattern of a defined population in a particular time and place. Although growth pattern is highly influenced by environmental conditions during childhood such as nutrition and infection, genetics specificities

27 and secular trends may also have an impact on growth observed in populations. Indeed, genetics determinants may influence height, timing of puberty, and in a lesser extent, weight, fat mass and fat distribution 153 . Therefore one may question whether an international growth standard for SAC and adolescents would be representative for specific populations. The need for the development of a new international growth standard of SAC and adolescents was raised because of the outbreak of childhood obesity and the release of a new international growth standard for children <5 y. Indeed the previous reference NCHS/WHO reference, the CDC2000 reference and the IOTF cutoffs are skewed which leads to underestimate obesity in SAC and adolescents 154 . Even if decades ago it was admitted that growth patterns were similar among different subpopulations for children 5 y when they are exposed to the same external growth factors 155 , the difference of the genetic impact on growth, maturation and puberty during school age and adolescence across populations remains unknown and needs to be investigated. In research comparing children, even healthy and privileged, from Africa, East Asia, South Asia, West Asia, and Europe to the NCHS/WHO reference, it was observed that secular trends in linear growth still may be occurring in some of the regions. Although African children and adolescents of upper socioeconomic status and African American achieved similar heights to the NCHS/WHO reference, children from Asia had lower heights. Finally heights of these privileged children vary less than 4cm from 7y to the beginning of the puberty, and afterwards they are around 5-6cm below NCHS/WHO reference (except for children from Northern Europe which are above the reference) 156 . These findings challenge the legitimacy of applying a growth reference based on a single population and of combining subpopulations to establish an international growth standard given the possible genetic differences in growth potential. Therefore, the sampling method for the development of an international growth standard for SAC and adolescents should at minima include multi ethnic sampling strategies designed to capture the variation in human growth patterns 154 . But facing the methodological heterogeneousness of datasets from various countries, WHO smoothed the 1977 NCHS/WHO reference using the original sample of well-nourished non-obese American children 157 which resulted in a new international reference of Height-for-Age and BMI-for-Age for children and adolescents aged 5-19 y. One may question the applicability of this reference to children and adolescents from developing countries. Moreover, BMI is related to lean mass and fat mass but it is not able to differentiate between them, so BMI is limited to evaluate body composition, which give precious information to assess nutritional status. Assessing obesity and eating disorders using weight for height or BMI gives poor sensibility for monitoring response to treatment. Knowledge on the distribution of body fat data is needed to prevent and diagnose cardiovascular disease hypertension and type 2 diabetes which are known to start during childhood and adolescence 30 . In LMIC countries, data of body composition are useful to understand the tissue accretion patterns among malnourished children as well as the risks for chronic disease in populations 30 . The “4 - component model” is suggested by Wells to be the gold standard to measure body composition 30 . However, it consists in collecting data on body weight, body volume, total body water, and bone mineral mass, which is expensive and complicated in field epidemiological surveys. Cheap and easy measurements such as skinfold thicknesses and circumferences were used to estimate calorie reserve in the form of fat, and protein reserve in the form of muscle 158,159 . However, their use to evaluate undernourishment and body composition among children and adolescents from developing countries is limited. First, only few data of TSF and MUAC or derived indicators are available as reference for SAC and adolescents of LMIC. In the past, Frisancho developed reference for MUAC, TSF and fat area for SAC and adolescent, by gender

28 and age range of one year. He also established reference for muscle area by gender and height. However, these references were based on white individuals from the NCHS reference. Some references were established in only a few countries 160,161 . Secondly, skinfolds thicknesses do not reflect the total amount of fat in the body, thus data on skinfold thickness may poorly predict lean mass 162 . Between-population variability in body composition may be partly influenced by genetic factors and therefore a universal body composition standard may not be possible or relevant 30 . A study on infants born to European, Indian, and mixed couples living in UK revealed that both fathers and mothers contributed to ethnic differences in fetal growth and birth weight, which suggest genetic influences or a long-term nutritional consequence over generations 163 . Consequently, the choice of the methodology to collect data and the absence of consensus about body composition patterns are 2 main issues in the assessment of body composition of children and adolescents from LMIC. Therefore the new international reference of WHO 2007 of height for age and BMI for age is challenged as an international standard.

The example of Cambodian SAC vs. Senegalese SAC In this thesis, anthropometric status of SAC from Cambodian and Senegalese samples according to various references is presented in table 2. Height and weight were lower in Cambodian children, which were younger than Senegalese children. Cambodian children were more stunted than Senegalese children. HAZ, BAZ, WAZ were significantly lower in Cambodian children. Undernourishment was significantly more prevalent in Cambodian children compared to Senegalese children according to prevalence of low BMI, low BMI, low TSF, low MUAC, low muscle area, low muscle circumference and low fat area. However, ratio of muscle-arm area and ratio muscle-fat area were significantly higher in Cambodian children than in Senegalese children, suggesting relatively more lean mass and less fat mass. These findings may result from different nutritional past in the 2 populations, secular trends, or genetic determinants.

1.4.1 Micronutrient indicators

WHO recommends indicators to assess folate, iron, vitamin A and iodine deficiency in populations (table 3). However, only for iodine deficiency cut-offs were established based on populations of SAC. No official indicator of other micronutrient status are available, therefore cut-offs and indicators used in SAC and adolescents vary across studies.

TABLE 2 EVALUATION OF ANTHROPOMETRIC STATUS OF CAMBODIAN AND SENEGALESE SCHOOL -AGED CHILDREN AND ADOLESCENTS

N Mean value SD N Mean value SD Cut-offs and interpretation Prevalence (%) P-value* P-value** Range Reference Cambodia Senegal Cambodia Senegal

Height (cm) 2549 124.5 11.4 602 138.7 14.3 <0.001 -

<0.001 <-3 z-scores : severe stunting 12.3 0.7 NCHS (National Center Height for Age (z- 2549 -1.8 1.1 602 -0.1 1.2 <0.001 5-19y for Health Statistics) scores) <0.001 [-3,-2[ z-scores : moderate stunting 29.1 4.2 1977

Weight (kg) 2550 22.7 5.6 602 30.1 9.6 <0.001 - -

1483 -1.8 0.9 292 -0.6 1.2 <0.001 <-3 z-scores : severe wasting 11.4 1.0 NCHS (National Center Weight for Age <0.001 5-10y for Health Statistics) (z-scores) [-3,-2[ z-scores : moderate wasting 30.3 5.8 1977

BMI corresponding to adult BMI <16 5.9 6.5 :thinness grade 3 pooled national surveys (Brazil, Great BMI corresponding to adult BMI BMI (kg/m 2) 2549 14.4 1.3 602 15.2 2.3 <0.001 15 9.1 <0.001 2 - 18y Britain, Hong Kong the [16;17.5[: thinness grade 2 Netherlands, BMI corresponding to adult BMI Singapore, USA) 39.5 32.7 [17;18.5[ :thinness grade 1

<-3 z-scores : severe thinness 5.0 5.8

[-3,-2[ z-scores : moderate thinness 20.6 12.8 NCHS (National Center BMI for Age (z- 2541 -1.5 0.9 602 -1.1 1.1 <0.001 <0.001 5-19y for Health Statistics) scores) ]1;2] z-scores : overweight 0.1 0.0 1977

>2 z-scores : obesity 0.0 0.3

* T-test ** χ2 test

TABLE 2 (CONTINUED ) EVALUATION OF ANTHROPOMETRIC STATUS OF CAMBODIAN AND SENEGALESE SCHOOL -AGED CHILDREN AND ADOLESCENTS

N Mean value SD N Mean value SD Cut-offs and interpretation Prevalence (%) P-value* P-value** Range Reference Cambodia Senegal Cambodia Senegal

MUAC (cm) 2550 16.8 1.7 601 18.5 2.4 <0.001 <5th percentile (for age range 1 y) 66.6 38.8 <0.001 1-75 y

<5th percentile (for age range 1 y) 47.1 32.9 <0.001 0-44 y

TSF (mm) 2549 6.4 2.0 602 7.6 3.2 <0.001 <9mm (girls), 7mm (boys) : underweight (corresponding to 15% 58.4 44.9 <0.001 10-17 y fat and 7%fat) US health and nut <5th percentile (for age range 1 y) 43.1 21.1 <0.001 1-75 y examination survey Muscle area 1971-74 - whites 2549 1770 367 601 2096 500 <0.001 (mm 2) <5th percentile (for height range 2 84-180 18.6 25.8 <0.001 cm) cm

Muscle circumference 2549 148 15 601 161 19 <5th percentile (for age range 1 y) 43.0 20.8 <0.001 1-75 y (mm)

Fat area (mm2) 2549 513 196 601 669 355 <0.001 <5th percentile (for age range 1 y) 61.0 41.8 <0.001 1-75 y

Proportion muscle / arm 2549 77.7 5.4 601.0 76.7 6.1 <0.001 area (%) Ratio muscle area/fat area 2549 3.7 1.1 601 3.5 1.1 <0.001 (%)

Age (years) 2549 9.5 2.4 602 10.2 2.4 <0.001

* T-test ** χ2 test

TABLE 3 INDICATORS OF MICRONU TRIENT STATUS AT POP ULATION LEVELS AS DE FINED BY WHO

Micronutrient Indicator Cut-off Interpretation Conditions Based on apply on

≥5 y except pregnant < 15 ug/L depleted iron stores known as based on studies examining women) sensitive to ferritin and microcytic anemia, ferritin < 30 ug/L depleted iron stores in the presence of inflamation inflammation (cf or response to iron, or bone <5 y Chapter 5.1 ) marrow Iron > 200 ug/L severe risk of iron overload adults observed as transferrin sensitive to depends upon the commercial > 5 - 8 mg /L tissue iron deficiency - soluble receptor inflammation (cf assay used Chapter 5.1 ) median < 20 ug/L insufficient iodine intake, severe iodine deficiency insufficient iodine intake, moderate iodine median 20-49 ug/L deficiency median 50-99 ug/L insufficient iodine intake, mild iodine deficiency all age groups, except single urine studies on populations of SAC Iodine urinary iodine pregnant and lactating sample (≥6 y) median 100-199 ug/L adequate iodine intake and nutrition women iodine intake above requirements, risk of more median 200-299 ug/L than adequate iodine intake excessive iorine intake, risk of iodine-induced ≥ 300 ug/L hyperthyroidism, autoimmune thyroid disease serum/plasma Vitamin A <0.35 umol/L severe vitamin A deficiency 6-71 months retinol 0.35-0.7 umol/L moderate vitamin A deficiency 24-71 months

all age groups, except pregnant women and serum/plasma based on homocysteine WRA (in which red Folate folate levels < 4 ng/mL concentrations as metabolic folate deficiency blood cell folate level indicator, in american men and < 400 ng/mL indicates women aged 30y and older folate insufficient to red blood cell prevent NTDs) folate level < 151 ng/mL

1.5. Review of coverage and cost-effectiveness of school feeding programs and staple food fortification: how to improve nutrition among school- aged children and adolescents?

1.5.1 School feeding programs (SMP)

1.5.1.1 Progress of coverage of school feeding program

WFP evaluated the worldwide situation of school feeding in 2013 through the analysis of 169 countries. 368 millions of children receive a meal every school day, of which 30 million in Africa and 168 million in SEA and Pacific. 25 million of these daily meals are distributed by WFP. In low income countries only 18% of children receive free school meals, as school feeding is usually only available in some areas selected for their vulnerability. 29 . Between 2008 and 2013, 39 countries have scaled-up school feeding in response to a crisis with respectively 13, 12, 8, and 5 countries regarding to the food crisis, an armed conflict, a natural disaster or the financial crisis respectively 29 . For example, the number of beneficiaries in Haiti has been tripled following the recent earthquake. In Togo, responding to the food crisis, 20,000 children living in food-deprived rural areas started in 2010 to be part of the school feeding program 164 . School feeding consists in providing school meals or snacks to be eaten during school hours or in distributing dry take home food rations to children at the end of each month if they attended school regularly 29 .

1.5.1.2 Benefits: decreasing hunger, incentive for school attendance, improvement of nutritional status

Research shows that providing school meals, snacks, and take-home rations through school feeding programs can alleviate short-term hunger, increase children’s abilities to concentrate, learn, perform specific tasks, and school meals have been linked to an increase in the enrolment of girls. These effects seem to be greater among children who are also chronically undernourished, usually the poorest children 164,165 . Attendance is increased by 2 to 17 days per year depending on studies 165 . For example, improvement of cognition scores was observed in Jamaica, with a stronger effect in undernourished children 166 , and in Kenya 20 . Math scores, physical activity, behaviors, and muscle mass were also improved in Kenyan children as well as height gain in those who were stunted 20 . School feeding even had a beneficial impact on weight gain in younger siblings of beneficiaries of take home rations in Burkina Faso 167 . In Bangladesh, the implementation of school meals increased the attendance by 6% and the enrollment by 14% 11 . Therefore, school feeding interventions can have a positive impact on health and schooling outcomes, even if the food is not fortified. Local food supply for school feeding, often known as “Home Grown School Feeding” (HGSF), can be a strategy t o offer more dietary diversity and fresh products 29 . HGSF programs conducted in Brazil and Chili 29 were successful to stimulate the economy, which could have long-term beneficial effects on reduction of poverty and nutrition in the community.

1.5.1.3 Costs of school feeding

According to WFP, school feeding worldwide represents an annual investment of 75 million USD. In low income countries, school feeding is estimated by WFP to cost on average 56 USD per child per year 29 . A cost-effectiveness study estimated the costs and the benefits of unfortified school meals in 6 countries of Sub-Saharan Africa 165 . Costs were standardized for 200-day school year and a 700 kcal ratio and took into account pipeline breaks. It varied between 28 and 68 USD per child and per year. The cost of school feeding varies according to location, geography, type and amount of commodities, number of children. Cost is higher for example in landlocked countries due to higher transport costs. A larger number of children allow economies of scale. Commodities accounted for more than half of the costs. The rest consists in transport, storage, handling and direct operational costs. Cost of school feeding per outcome were estimated in the review : 18 USD per IQ point, 33 USD per point on math achievement or aptitude, 2-8 USD per extra day of attendance, 27-160 USD per additional cm of height and 55- 160 USD per additional kg of weight. Effects appear significant but small, but they may have been underestimated. Effect of school feeding program on IQ, weight gain and attendance are similar to those of other school health and nutrition interventions, such as iron supplementation, deworming or malaria prevention, which cost less than 4USD per child and per year. However, these costs may not have been calculated as accurately as the cost of school feeding program.

1.5.1.4 Challenges and perspectives of school feeding in developing countries

Given the drastic increase of school meals distribution in the past decade 29 , the challenge is not any longer to demonstrate the need to implement school feeding programs but to determine how to improve the efficiency of these programs. Food basket, size of rations and nutritional value of meals are highly variable among countries 29 . Tracks to diminish costs and improve the health benefits of school feeding program should be explored. Distributing fortified biscuits, which cost 11-12 USD per child per year, may reduce the cost as well as the substitution effect (children less fed at home when they are fed at school) 165 . Targeting undernourished children, according to areas, the times of year when poverty is highest and the attendance lowest would increase the effect and reduce the cost of feeding all children all year along. Home grown school feeding program (HGSF) is a new concept and only a few countries implement it on a national scale. However, these programs are becoming popular among governments, partners and donors, giving the benefits of HGSF on local production and increasing availability of funding for school feeding 168 . Local procurement of food is a possible means to achieve sustainable programs, to reduce costs and at the same time to stimulate local agricultural production 164,165 . It is critical that long-term sustainability is incorporated into programs from their inception, and that programs are continuously revisited as they evolve. As more and more governments seek to expand these programs in their countries, it is important to have more opportunities for knowledge sharing among developing countries that focus on ways to improve the procurement of locally available nutritious foods and compare best practices 164 . Combining SFP with other interventions such as micronutrient fortification or deworming is another strategy to enlarge the impact on health and schooling outcomes 165 . School feeding program can integrate food fortification,

34 deworming, nutrition education and access to clean drinking water. Galloway et al recommend to implement accurate cost-effectiveness studies of school health and nutrition interventions to identify the most cost-effective package of interventions 165 .

1.5.2 Rice fortification, a promising strategy to prevent micronutrient deficiencies

1.5.2.1 Why choosing rice as a vehicle for fortification?

Food fortification such as iodization of salt, fortification of milk with vitamin D, and enrichment of flour and bread with vitamins B and iron has significantly contributed to the disappearance of goiter, rickets, beriberi, and pellagra, respectively caused by iodine, vitamin D, vitamin B1, and vitamin B3 deficiency in the United States since the early 20 Th century 169,170 . More recent trials have demonstrated beneficial impact of fortified foods on nutritional outcomes : in Vietnam, fortified fish sauce improved iron status and reduced anemia, fortification of monosodium glutamate in the Philippines, of margarine in Denmark the Philippines, and of sugar in Central America diminished vitamin A deficiency prevalence, and fortified milk reduced vitamin D deficiency in US 171,172 . Micronutrient deficiencies are associated with polished rice consumption in many developing countries. Although paddy rice is a natural good source of thiamine, riboflavin and niacin, hulling and milling turn polished rice into a poor source of micronutrients. Rice represents one to two thirds of caloric intake in 17 countries of Asia and the Pacific and in 6 Sub-Saharan countries, indicating diets lacking in diversity. Prevalence of anemia, stunting and thiamin, riboflavin and vitamin A deficiencies are generally higher in populations where polished rice is consumed 173 .

1.5.2.2 Current knowledge about the efficiency of fortified rice

Gains on health and schooling outcomes of fortified rice interventions among SAC were reported by several studies (table 4). Plus, compared to other interventions aiming at increasing micronutrient intake, food fortification is cheaper and has a high cost-benefit ratio 174 .

TABLE 4 EFFECTIVENESS OF RICE FORTIFICATION AMONG SCHOOL CHILDREN IN DEVELOPING COUNTRIES

Country / Sample Study method Study results Fortified rice Year

India 140 children 5- double blind placebo improvement of Extruded Iron Ultra Rice 11 participating controlled trial 8 ferritin 8.2 ± 2.1 µg/L (9.6 mg FE /g blended 2007- into the MD meal months (166 days of in FR-MDM group vs. with natural rice 1.6:98.4 2008 175 program (at BL, consumption) FR- decrease in control blend ratio around 40% of MDM (fortified rice- group (P<0.001). anemia, 30 % ID, Midday Meal) vs. decr ease of ID corresponded to 19mg 15% of IDA in UFR -MDM prevalence ( -19% in iron/d in MFPP both groups) unfortified rice- FR-MDM group (Micronized ferric Midday meal P<0.05) increase in phyrophosphate) control group +6% 15 mg FE /100 g blended raw rice

meant to meet the gap to fulfill iron RDA for Indian children

India 184 children double blind larger improvement Extruded iron-fortified with depleted randomized placebo of ferritin +9.5 µg/L rice (10 mg FE/g as MGFP 176 2006 iron stores 6-13 controlled trial 7 in iron group vs. +2.3 1:50 blend ratio y participating months iron µg /L in control group corresponding to 20 mg into a school fortified rice vs. non (P<0.05) iron/d in MGFP lunch program fortified rice in lunch (Micronized ground ferric meal decrease of ID/IDA pyrophosphate) prevalence -53%/- 15% in iron group vs. -30%/ -1% in control group (P<0.001) 200 -300 g cooked rice daily no effect on anthropometry and 20 mg FE /100 g raw morbidity blended r ice

Country / Sample Study method Study results Fortified rice Year

Thailand 203 children double blind Larger improvement Hot-extruded fortified with low zinc randomized of serum Zn in the rice (10mg FE, 9 mg Zn, 177 2008 status 7-12 controlled placebo fortified rice group 1050 µg VA/g, 1:50 blend receiving lunch trial 5 months iron- +11.3 µmol/L vs. ratio) meals and daily vitamin A-zinc +10.6 in the control milk (but not fortified rice vs. non group (P<005). No assumption : increase of during the fortified rice in lunch effect on FE or VA intake by 97.5%, 10% of summer before meal status (not deficient) bioavailability of Fe, the study) moderate bioavailability of Zinc, 40% of loss of vitamin A

20 mg FE, 18 mg zn, 2100 µg VA /100 g or raw blended rice

India 258 children 6- double blind in high-iron and low DSM fortified rice 12 y (prevalence randomized placebo iron fortified rice 2009- of ID, VAD and controlled trial 6 group VS control, low -iron rice 6.25 mg 130 2010 ZnD<10%) months low-iron larger improvement FE/ 100 g blended raw fortified rice vs. high of plasma vitamin rice iron fortified rice vs. B12 (+92 pmol/L and high iron rice 12.5 mg unfortified rice in +105 pmol/L vs. 20 FE/100 g blended raw unch meal pmol/L P<0.001), rice plasma homocystein (-2.6 and -2.3 µmol/L Micronized ground ferric vs. 0 µmol/L P<0.05 pyrophosphate and physical performance Assumptions: to provide (respectively – no 40-50% of RNI for vitamin effect on iron, zinc, A, thumaine, niacin, vitamin A status, or vitamin B6, folate, iron, cognition) zinc for 7-9y

1.5.2.3 Technologies to fortify rice

The four technologies currently available to produce fortified kernels assigned to be blend to unfortified rice are described in table 5.

TABLE 5 TECHNOLOGIES TO PRODUCE FORTIFIED RICE KERNELS 178

Technology Description

Dough made of rice flour, the micronutrient mix and water passes hot extrusion through an extruder which cut it into grain like structures, at high temperatures (70-110°C)

Similarly, rice-shaped kernels are produced through an extruder, but cold extrusion at temperatures below 70°C. Cold-extruded kernels are easier to distinguish from real rice than hot extruded kernels

The fortificant is mixed with waxes and gum, and then sprayed on the coating surface of rice kernels in several layers,, which are afterwards mixed with unfortified rice

Similarly to flour fortification, micronutrient powder is simply dusted dusting on rice grains and sticks to it because of electrostatis forces.

The choice of the technology depends on investment capacities and the level of consumer preferences : the fortified rice can be homogenous or on the contrary, fortified rice can be colored to be distinct from non-fortified rice, indicating higher nutritional value 178 . Dusting technology is cheap as no blending is needed, but it is unsuitable for developing countries where rice is washed before cooking. Coated rice usually leads to changes in color, smell and taste. The extrusion technology is considered as the most adapted to areas where populations wash or rinse the rice and it produces acceptable rice for taste and appearance 178 . Several studies had demonstrated the acceptability of extruded fortified rice 179 ,176 ,175 . With hot extrusion, kernels are almost undistinguishable from natural rice (transparency, sheen, consistency, and flavor) while cold extrusion produces more opaque kernels. However, hot extrusion implies high investment costs therefore coating and cold extrusion are more appropriate for small or pilot programs while hot extrusion is more suitable for long-term rice fortification programs 178 .

1.5.2.4 Selection of fortificants and fortification levels regarding to desired impact and toxicity

It is recommended to choose a condiment or staple food as vehicles for food fortification. Condiments are usually fortified with only one or 2 nutrients while several minerals and vitamins can be added to cereals without modifying the taste and the color. Combining iron- fortification of 2 foods may be useful to minimize the alteration of taste and color 173 . Vitamin A and iron deficiencies are widely spread in the developing world and were shown to be reduced

38 using fortification and supplementation. For example, flour fortification with folic acid drastically diminished the incidence of NTDs in the USA. In addition, zinc supplementation can reduce mortality and diarrhea prevalence and might improve growth. Vitamin B12 is only present in animal source foods which are usually hardly consumed in non-affluent populations due to the relative high price. Therefore iron, vitamin A, folic acid, zinc and vitamin B12 are recommended as rice fortificants 180 . Recent studies revealed insufficient intake of vitamin B6, niacin and thiamin in several Asian subpopulations. Also, beriberi was shown to be prevalent in SEA. Therefore, thiamin, niacin and vitamin B6 are recommend additions too 180 . Vitamin B2 (riboflavin), vitamin C, iron as ferrous sulfate and beta carotene can change the color of the fortified rice kernels 173 , perhaps reducing acceptability. Currently, ferric pyrophosphate is the preferred form of iron as fortificant in rice, although bioavailability is lower than from for example iron sulphate 176,178 . Retinyl esters are an alternative form to vitamin A which does not affect the color of fortified kernels. Dry forms of vitamin E, vitamin B1, niacin, folic acid, vitamin B12, zinc as zinc oxide and selenium are all compatible with rice fortification. Vitamin K, biotin, pantothenic acid, vitamin B6 can also be added to rice without any impact on taste or color 173 . Ideally, the WHO/FAO recommend to set fortification level following : measurement of dietary micronutrient intake among populations at risk for deficiency followed by estimation of the bioavailability of zinc and iron in the local diet in order to compare the absorbed amount to requirements 171 . But such data are often unavailable in developing countries. Therefore, based on flour fortification level and on the same rationale linked to rice fortification, de Pee proposed recommended levels (mg /100g) to achieve intake that meets the estimated average requirements of adults 180 (table 6).

TABLE 6 RECOMMENDED NUTRIENTS AND NUTRIENT LEVELS (MG /100 G OF RICE ) FOR RICE FORTIFICATION WHERE RICE CONSUMPTION IS 150-300 G PER CAPITA PER DAY BY DE PEE

Vitamin Vitamin Iron Folic acid Vitamin A Zinc Thiamin Niacin B12 B6

7 0.13 0.001 0.15 6 0.5 7 0.6

Bioavailability of iron and zinc related to phytate content of polished rice were accounted. Folic acid fortification level is established to meet the RNI for anyone older than 14 y while levels for vitamin B12, B6 and B3 meets the EAR 180 . Toxicity level of some micronutrients should be carefully taken into account when fortificant levels are established. Vitamin A toxicity can result in liver toxicity, loss of appetite ophthalmological and dermatological disorders. Vitamin D toxicity is associated with hypocalcaemia and associated disorders. Excessive iron results in gastrointestinal damage, liver necrosis and formation of free radicals 181 . Consequently, iron, vitamin A (retinol) and, vitamin D, are classified as nutrients with a small safety margin, with upper intake levels less than 5 times superior to the recommended intake for which fortification level should be carefully considered 181 . Indeed, the RNI for vitamin A is close to upper intake, which raises concern when considering vitamin A fortification, so recommended levels correspond to 25% of the RNI of women preschool and SAC 180 . This limitation is questioned as it was set for affluent populations and could conflict with the need to increase intake in less developed countries. Furthermore, vitamin A capsules distributed to young children contain a much higher dose that the upper intake level. Therefore vitamin A level could be increased 182 . In order to stay below the upper intake level for niacin of 35 mg/day, recommendation for niacin level should be carefully adapted to the prevalence of inadequate

39 intake and the consumption range of fortified rice. As for vitamin A, much higher doses of niacin, up to several grams/d are given as medication to lower LDL concentrations. Vitamin E, B6, B12 and C are classified in intermediate category, while vitamin K, thiamine, riboflavin, pantothenic and biotin are harmless even at 100 times the recommendation 181 There is no upper limit for vitamin B12, thiamin and vitamin B6. However, because levels of fortification are determined as function of the dietary needs of healthy adults, vulnerable subgroups with high requirements such as pregnant and lactating women and infants may need additional specific interventions. In the case of SAC, fortification for adolescent girls at risk to be pregnant may not be sufficient to meet their high nutrient requirements. For example Iron-folate tablets will be needed for pregnant women.

1.5.2.5 Costs of rice fortification programs

Implementing fortification of rice may be useful if the population consume at least 100g of rice daily (36 kg/year) 178 . Costs linked to production and distribution of fortified rice consists in production and transportation of fortified kernels, in blending of fortified kernels with unfortified rice, and in distribution or sales of rice 174. However, costs associated to pre and post-implementation research, and costs to support and monitor the rice fortification program have to be taken into account too.

Roks analyzed the cost of rice fortification at each step 174 . Installing respectively a hot or a cold extrusion facility cost around $4 million (which covers 1.5 to 8 million people, depending on annual consumption) and $0.5 million (which covers 0.3 to 2.5 million people) respectively, and start to be profitable from an estimated yearly demand of 1500 MT or 300 MT of fortified kernels 178 respectively. The method to dry extruded fortified kernels also has an impact on the overall cost: natural dry may be cheaper than using a dryer system but it can affect the quality and hygiene of kernels 174 . The cost of investment can be supported by milling industry, e.g based on perceived market potential or corporate social responsibility, but external support may be needed too. Broken rice is a by-product of the rice milling industry and can be used to produce rice powder intended to be mixed with micronutrients to produce fortified rice kernels. Thus, fortification of rice allows to add value to broken rice 174 . Cost to produce fortified kernels vary according to several parameters 174 . A lower mixing ratio associated with higher content of micronutrient in kernels would be more cost effective but could be a problem for taste or look of fortified rice. Blended ratio is usually between 1:200 (0.5%) and 1:50 (2%) 174 . The cost usually decreases when the order size increase. Rice fortification is practical to implement in mills who product more than 5 MT per hour or 15000 MT / year 174 . The overall cost of production is $1.89 to $6/kg of fortified kernels, 2-2.5 /kg on average, of which ~50% ($1 - $1.4) is allocated to the actual production of the fortified kernel 174 . The higher overall cost of rice fortification compared to other food fortification is associated with the synthesis and blending of artificial kernels, thus the price of the micronutrient premix has only a small impact on the cost of fortified kernels production (26% to 33%), and hence a small impact on the overall cost of rice fortification program. Therefore the choice of micronutrients and their amount is supposed not to be limited for economic reasons, but to be guided also by nutrient requirements and safety (see above) 174 . As an example, if producing fortified kernels cost 3USD/kg (of which 1 USD/kg represents the fortificant premix cost), the blending ratio is 1:100, and the price of normal rice is USD 400/MT, the cost to fortify 1MT of rice will be:

400*99%+3000*1%=426 USD, therefore fortified rice will be 6% more expensive than unfortified rice. For populations consuming between 100 and 300 g of uncooked rice per day, the additional cost of fortified rice compared to unfortified rice is estimated to be between $0.22 and $2.18 per year and per person 178 . For example, a fortified rice providing more than 100% of EAR of vitamin B12, folic acid, vitamin B1, between 50% an 76% of vitamin A, niacin and zinc EAR, and 18% of iron EAR to women consuming 200 g of rice per day would only cost an additional 0.22$ per woman per year 178 . This only includes the price to produce fortified kernels, not the additional program costs. Overall, more research about costs of rice fortification is needed. The fortified kernels cost could be supported by the consumer in commercial settings. When distributed, the additional cost could be covered by the party who operates the existing rice distribution, as national government or WFP, supported by donors. Fortified rice expires one year after blending, due to decrease micronutrient content and product acceptability rather than harmful effects 174 . To our knowledge, facilities to produce hot or cold extruded fortified kernels are only available in Asia (China, India, the Philippines) and in Latin America (Colombia, the Dominican Republic, Brazil, Costa Rica). 174 . Consequently, the cost of transportation of fortified rice kernels have to account transportation from abroad to the country, which vary according to the mode of transport, quantity and market price, and importation fees and customs clearance fees. Then, costs of transportation inside the country depend on distance, complexity of the supply chain. In order to blend fortified kernels with unfortified rice, an accurate proportioning device has to be added at the end of the processing line in sophisticated rice milling facilities, which usually cost up to $20,000, but which will not need extra labor cost once installed. Otherwise simple manual mixers cost approximately $200 but need extra labor cost 174 . If the rice milling is centralized there will be less logistic difficulties and the cost efficiency of rice fortification would be higher 174 . Fortified food are meant to be distributed through the existing sales or distribution tracks for unfortified food, therefore no additional cost related to distribution or sales should be input to food fortification. However, in case fortified rice is to sell, social marketing expenses may be needed 174 . Scale-up the distribution of fortify rice will allow economy scales on fortified kernels production and transportation costs.

1.5.2.6 Key success factors and challenges of rice fortification

To ensure efficiency of rice fortification, sensory properties of rice as taste, smell and texture should be similar to those of unfortified rice, fortificants have to be stable and bioavailable, fortification has to be regulated by authorities and accepted by the target population 174 . Keys success factors for implementation of rice fortification were identified by Piccoli et al 173 : - adequate per capita rice consumption : 70-100 g uncooked fortified rice/ person/day or 100-200 g cooker fortified rice /person /day - high level of micronutrient deficiencies - Existence of government-sponsored or manager safety nets: India, Indonesia and Bangladesh implemented public safety net programs distributing free or highly subsidized rice, targeting low income households, adolescent girls, under 5 years old children, pregnant and lactating women or SAC. When it is possible, WFP use fortified food in its program in disaster-stricken areas and its school feeding program. - Large market size and awareness of consumers for market-driven fortification: If consumers are aware of the benefits of fortified rice, they may be willing to pay the

small additional price linked to rice fortification, which will be as reduced as the market is large, due to scale economies. - Consumer awareness and/or acceptance: as rice is a staple food in many countries, the type of rice with specific shine, color or taste can have social connotations rooted in traditions. Therefore, if they are aware of the benefits of fortification, they may be willing to pay for and to consume fortified rice, which is not likely to happen among poor and uneducated populations. This could be addressed by safety net or subsidy strategies. - Adapted technologies: acceptance etc. In countries were soaking and washing are practiced, coating and dusting technologies are not relevant - adequate capacity in the rice processing industry : fortification of rice programs would be more effective if there are a few large mills with appropriate technology rather than many smaller mills - No or limited restrictions on rice trade : subsidies to increase fortified rice consumption and production, and the reduction of import barriers for fortified rice kernels - ease of doing business : rice fortification will be more easily implemented in countries less affected by corruption and bureaucracy

To conclude, school feeding and rice fortification are promising nutrition interventions likely to improve nutritional status of SAC in developing countries. School feeding is being widely implemented in the world, including developing countries, in order to improve school attendance and concentration but cost-effectiveness and nutritional benefits could be improved through food fortification. Fortification of rice is promising for bridging the nutrient gap in poor settings. Combining rice fortification with school feeding program could be a cost-effective way to increase micronutrient intake. Effects on the use of rice fortified with multiple micronutrients on micronutrient status and cognitive performance are reported in this thesis in 2 different papers (chapter 4).

Chapter 2. Nutritional status and its dietary determinants of urban African school-aged children and adolescents: Case study in Senegal

2.1. Anthropometric and micronutrient status of school-children in an urban West Africa setting: a cross-sectional study in Dakar, Senegal (Published paper)

Published in Plos One (2013) Marion Fiorentino a, Guillaume Bastard b, Malick Sembène c, Sonia Fortin a, Pierre Traissac a, Edwige Landais a, Christèle Icard-Vernière a, Frank T Wieringa a, Jacques Berger a

Introduction

The prevalence of food insecurity in Sub-Saharan Africa is the highest in the world, with rates as high as 30% of the population being undernourished 183 . For instance, 26% of the population in Senegal is undernourished, which ranks it 155 th over 187 countries in the 2011 Human Development Index 184 . Due to considerable rural migration and urbanization during the last decades, 42% of the Senegalese population now live in urban areas 185 . Many households in the Dakar area are therefore without basic infrastructures, while simultaneously vulnerable to food insecurity 186 . Moreover, albeit the general assumption that urban populations have access to more diversified foods, studies in West Africa have shown that micronutrient status can be low in urban areas 187 .

Nutrition interventions generally neglect school children despite their high prevalence of malnutrition and micronutrient deficiency 17 . Iodine, iron and folic acid micronutrient deficiencies affect the development of the brain and cognitive functions of school children 36 . Iodine deficiency, even mild, could impede full intellectual potential 188 with differences in intellect as large as 10 – 15% between iodine deficient and non-deficient populations. On the other hand, deficiencies of vitamin A and zinc are associated with different scenarios affecting school performance, such as absenteeism due to increased morbidity 189,190 .

Data on nutritional status of school children in Senegal are scare and recent data on school children living in urban areas are lacking. A study in 1994 conducted on a representative sample of 774 children (aged from 5 to 15 y) of state primary schools of the Dakar department showed that 34% of pupils were anemic, 10% underweight, 5% stunted and 11% wasted 191 More recent, a study on Senegalese food practices and nutrition was conducted to identify strategies to reduce malnutrition 145 . New food practices are emerging in urban areas with food prepared at home decreasing, while street foods purchases increased. Furthermore, when a child reaches school age, it can be observed that mothers take less care of their children’s diet. The rarity of school canteens in urban state school further reduces meal opportunities for children 145 While school children consume less meals with adults, an increasing number of households can’t afford to spend money on snacks for their children. Many children declare having difficulties to concentrate in class due to hunger 186 .

It is therefore likely that the nutritional intake of foods from school-aged children in Senegalese urban areas is inadequate. The main objective of the study lies in the assessment of the anthropometric and micronutrient status of school children attending state primary schools in Dakar and suburbs. The Dakar region was selected because it has the highest population density (4513 people/km2) and schooling rate (90%) of the country 192 .

Participants and methods

Study area and population

The Dakar region, including the capital Dakar and its suburbs, is located on the Cap-Vert peninsula. It represents 0.3% of the Senegal area and 21% of its population. The study was conducted in February-March 2010. The target population, children from state primary schools of Dakar region, was estimated to be over 200,000 children, distributed among 370 state schools.

Study design

The study was a cross-sectional survey. A two-stage cluster sampling method was chosen with schools considered as primary sampling units.Within randomly selected schools, and without criteria for age and gender, children were randomized as final sampling units. The required number of participants was calculated following the formula n = (1.64²×P× (1-P))/ m²; where the prevalence P was estimated equal to 0.5 and the expected precision m for this prevalence to be 0.05. Moreover, a design effect equal to 2 was chosen 193 leading to a required sampling size of 538 children. Thirty schools were randomly selected, with 20 children per school, giving a final sample size of 600 children. The regional education authority provided the list of schools, while children were randomly selected from lists provided by the school directors.

Data collection

Mothers or caretakers of children were surveyed at home for socioeconomic characteristics of the households. Birth dates of children were recorded from the school lists and checked with birth certificate or identity card. When official documents were mi ssing, children’s mothers or caretakers were questioned using a local events calendar. Children were defined as participants less than 10 y, and adolescents as participants from 10 y and above, according to the World Health Organization (WHO) definition 14 .

The nutritional assessment period lasted 6 weeks between February to March 2010, with one school visited each day. Blood, urine samples and anthropometric measurements were collected at schools in the morning between 8 and 10 AM. They were then verified for their identity. In order to define their fasting status, children were also asked the last time they consumed food. Weight and height were measured without footwear and wearing minimal clothes. To avoid between-measurers variability, all anthropometric measurements were performed by only one trained anthropometrist. The accuracy of the scale and the stadiometer was checked every day using a set of 2 calibration weights and one calibration tape. Height was measured twice to the nearest 0.1 cm on a Seca 214 stadiometer and mean values were used. When differences between two measures of height for the same child exceeded 0.5 cm, measurements were repeated. Weight was measured once to the nearest 100 g on a Pespe T125 Terraillon scale. Height-for-age (HAZ) and BMI-for-age z-scores (BAZ) were calculated according to the WHO 2006 reference 157 Stunting was defined by HAZ<-2 z-scores. Overweight was defined by BAZ between 1 z-scores and 2 z-scores and obesity by BAZ >2 z-scores. Two growth international references were used to classify thinness: the WHO reference defining mild, moderate and severe thinness respectively by z-scores between -2 and -1 , between -3 and -2 and <-3 157 ; the

IS reference defining grades 1, 2, 3 of thinness corresponding to WHO cut-off for BMI of 16, 17, and 18.5 respectively at age 18 194

Four (4) ml of venous blood were collected in a Terumo heparin Venosafe vacutainer with heparin by an experienced phlebotomist (using sterile single-use material). Urine was taken in a sterile container. Urine and blood samples were stored immediately in an icebox containing ice- packs and transported to the laboratory within a maximum of 3 hours after the first sample withdrawal.

Hemoglobin concentration (Hb) was measured at arrival at the laboratory of Pasteur Institute in Dakar on whole blood with HemoCue® Hb 201+ and HemoCue controls (Hemotrol low, medium, high, HemoCue®). Moreover 5% of blood samples were measured by hematology analyzer Cell-dyn® as an external control. Blood samples were centrifuged at 4000 x g for 10 minutes at -4°C, plasma aliquoted in 4 eppendorf tubes and stored at -20°C for 6 weeks until completion of the field work and sent with dry ice to the Nutripass laboratory of the Institut de Recherche pour le Développement (IRD, Montpellier, France) for zinc and iodine measurements and to the CBS laboratory (Willstaett, Germany) for determination of retinol-binding protein (RBP), C-reactive protein (CRP), ferritin (FER), soluble transferrin receptor (TFR ) and α1 -acid- glycoprotein (AGP). RBP, FER, TFR, CRP, AGP were measured by a sandwich enzyme-linked immunosorbent assay (ELISA) technique 195 . Plasma zinc was measured by flame atomic absorption spectrophotometry (AAS), using trace-elements free procedures and urinary iodine (UIC) was measured using an ammonium persulfate method 196

Inflammation was determined by elevated CRP (>5 mg/L) and/or elevated AGP (> 1 g/L) allowing differentiation between incubation phase (high CRP), convalescence phase (both AGP and CRP elevated) and late convalescence phase (elevated AGP only) 197 Anemia was defined by Hb below cut-offs depending on age and gender: 115 g/L for participants <12y; 120 g/L for adolescent s between 12 and 15 y and girls ≥15 y; 130 g/L for boys ≥15 y ) ; severe anemia was defined as Hb<70 g/L; depleted iron stores were defined by corrected FER <15 µg/L 198 . Correction factors of FER were 0.77, 0.53 and 0.75 for participants respectively in incubation, early convalescence, and late convalescence phases 197 . Iron tissue deficiency was defined by TFR>8.3 mg/L. Low FER and high TFR are both considered as indicators of iron deficiency (ID) 198 so ID was defined by iron stores depleted (low FER) and/or iron tissue deficiency (high TFR). Body iron was calculated according to the formula of Cook : body iron (mg/kg) = - (log (TFR/FER ratio)-2.8229)/0.1207 199 . “Body iron deficiency” was defined by body iron <0. Vitamin A status was measured by RBP concentration which reflects plasma retinol concentration because RBP occurs in a 1:1:1 complex with retinol and transthyretin 200 RBP concentrations were corrected in participants with inflammation by factors 1.15, 1.32, 1.12 respectively for incubation, early convalescence and late convalescence 201 . Vitamin A deficiency (VAD) was defined by corrected RBP <0.7 µmol/L and ≥0.35 µmol/L and severe VAD was defined by corrected RBP <0.35 µmol/L 200 . Marginal VAD was defined for corrected RBP values ≥0.7 µm ol/L and <1.05 µmol/L 202 . Zinc deficiency (ZnD) was defined by plasma zinc concentration <0.65 mg/L for participants <10y independently of their fasting status and for participants >10 y, cut-offs are 0.66 mg/L for non fasting girls, 0.70 mg/L for fasting girls and non fasting boys, and 0.74 mg/L for fasting boys 203 .

Iodine deficiency (IDD) was defined by a median UIC below 100 µg/L and/or a proportion of participants below 50 µg/L higher than 20%. Mild IDD was defined by a median UIC between 50 and 99 µg/l, moderate IDD by a median IUC between 20 and 49 µg/l and severe IDD by a median UIC below 20 µg/l, iodine nutrition above requirements by a median UIC between 200 and 299 µg/l, and excessive iodine nutrition by a median UIC equal or above 300 µg/l 124

Ethics

The protocol was approved by the ethical committee of the National Health Research of Senegal. The school directors informed parents of the selected children on the purpose and proceedings of the study. Written informed consent was obtained from all parents at the beginning of the study. Severe anemic participants received iron supplementation as treatment.

Data management and statistical analysis

Data entry, including quality checks and validation by double entry of questionnaires, was performed with EpiData version 3.1 (EpiData Association, Odense, Denmark). Data management and analyses were performed with the SAS software version 9.2 (SAS, V9.2; SAS Institute, Cary, NC). All analyses took into account characteristics of the cluster sampling design using the appropriate survey procedures of SAS. Categorical variables were expressed as percentages and standard errors of prevalence (surveyfreq procedure). Interval variables were expressed as arithmetic means and standard errors of means (surveymeans procedure), except ferritin and transferrin whose distributions were not normal and which were also expressed as geometric means and standard errors of means. Associations between prevalence and gender or age group were assessed by prevalence OR using logistic regression models (survey logistic procedure). Comparisons of means between gender and age groups were done through ANOVA (surveyreg procedure). Thus regression models included relevant cofounders (gender, age group or interaction according to models) to estimate adjusted ORs and differences.

Results

In total, 604 children aged from 5 - 17 years participated in the study, 317 girls (52.5%) and 287 boys (47.5%, Table 7). About half of the participants were adolescents (n=287, 47.6%) and others were children (n=317, 52.4%). Of the children, 26% were schooled in Dakar, 57% in nearby suburbs of Dakar (Pikine, Thiaroye, Guediawaye, KeurMassar) and 17% in far suburbs (Rufisque and surroundings). Of the children, 5.7% of the subjects had an elevated CRP and 10.6% an elevated AGP. Prevalences of incubation (high CRP and low AGP), convalescence (high CRP and high AGP), and late convalescences were (low CRP and high AGP) 1.5%, 4.2% and 6.4% respectively. Whereas half of the mothers had a job (54%), only 26% finished primary school. Most (93 %) households had electricity, but only 41% had a fridge.

Six hundred and four school children participated to the study and were measured for anthropometry, with 596 blood samples being obtained. 12 samples are missing, due to the refusal of 3 children and insufficient blood collection from 9 others. Hb was measured on all blood samples whereas CRP, AGP, FER, TFR were measured on 594 samples and Zn on only 584 due to insufficient blood volume. A total of six hundred urine samples were collected with only 4 children refusing. Due to the loss of labels during the transfer of samples, it was decided that 4 would not be measures for iodine.

TABLE 7ANTHROPOMETRIC AND BIOCHEMICAL STATUS OF PARTICIPANTS FOR ALL AND DISAGGREGATED FOR CHILDREN (<10 Y) AND ADOLESCENTS (≥10 Y)

All Children Adolescents

Mean / Mean / Mean / n SE* n SE* n SE* p Prevalence Prevalence Prevalence BMI (kg/cm 2) 604 15.23 0.12 287 14.51 0.14 317 15.87 0.15 <0,0001 Thinness grade 1 a (%) 33.6% 2.1% 30.7% 2.8% 36.3% 2.7% NS

Thinness grade 2 b (%) 10.4% 1.2% 10.8% 1.7% 10.1% 1.4% NS

Thinness grade 3 c (%) 6.5% 0.9% 5.9% 1.5% 6.9% 1.5% NS

BAZ 602 -1.14 0.05 286 -1.04 0.07 316 -1.22 0.06 NS Mild thinness d 36.9% 2.2% 34.6% 3.2% 316 38.9% 3.0% NS

Moderate thinness e (%) scores 12.8% 1.3% 12.2% 1.8% 13.3% 1.9% NS

Severe thinness f (%) 5.6% 0.8% 3.8% 1.3% 316 7.3% 1.5% NS

HAZ 595 -0.13 0.06 279 0.10 0.09 316 -0.33 0.07 0.02 Stunting g (%) 4.9% 0.9% 3.2% 1.1% 6.3% 1.5% NS

Plasma retinol (µmol/l) 594 1.14 0.01 279 1.09 0.01 315 1.18 0.01 0.001 Vitamin A deficiency h (%) 3.0% 0.8% 4.3% 1.2% 1.9% 0.9% NS

Vitamin A marginal status i (%) 35.9% 2.0% 45.5% 3.2% 27.3% 2.5% 0.00

Plasma zinc (µmol/l) 584 0.75 0.01 269 0.75 0.02 0.75 0.01 NS

Zinc deficiency j (%) 25.9% 3.7% 23.4% 4.4% 27.9% 3.9% NS

Iodine (µg/l) 600 146.67 7.09 283 145.33 8.57 317 147.87 7.19 NS Iodine < 20 µg/l (%) 1.3% 0.47% 0.7% 0.5% 1.9% 0.9% NS

Iodine ≥ 20 µg/l and <50 µg/l (%) 6.0% 1.7% 6.0% 2.2% 6.0% 1.9% NS

Iodine ≥50 µg/l and <100 µg/l (%) 25.5% 2.9% 26.5% 3.5% 24.6% 3.3% NS

Iodine ≥100 µg/l and <200 µg/l (%) 44.8% 2.8% 46.3% 3.4% 43.5% 3.4% NS

Iodine ≥200 µg/l and <300 µg/l (%) 17.0% 2.5% 15.2% 3.3% 18.6% 2.4% 0.07

Iodine ≥ 300 µg/l 5.3% 1.3% 5.3% 1.6% 5.4% 1.6% NS

Hb (g/l) 596 12.58 0.05 279 12.51 0.08 12.65 0.09 NS

Anemia k (%) 14.4% 1.5% 13.3% 1.9% 15.5% 2.3% 0.07

Ferritin (µg/l)** 594 24.86 0.64 22.81 0.64 315 25.05 0.64 NS

Low ferritin l (%) 21.4% 1.8% 23.3% 2.5% 19.7% 2.0% NS

Transferrin receptor (mg/l)** 8.76 0.50 7.94 0.50 7.67 0.50 NS

High transferrin receptor m (%) 33.3% 2.1% 35.5% 2.9% 31.4% 3.0% NS

Body iron (mg/l) 2.56 0.13 2.32 0.16 2.77 0.19 NS

Negative body iron deficiency (%) 17.7% 1.6% 20.4% 2.4% 15.2% 1.6% NS

ID according to FER and sTFR(%) 39.1% 2.4% 41.2% 3.1% 37.1% 3.3% NS

IDA according to FER, sTFR and Hb(%) 10.6% 1.4% 10.0% 1.8% 11.1% 2.0% 0.21

IDA according to BodyIron and Hb(%) 7.1% 1.1% 6.8% 1.4% 7.3% 1.5% NS

*standard error ; **geometric means NS non significant a consistent with WHO adult 17 ≤ BMI < 18.5 (IS reference); b consistent with WHO adult 16 ≤ BMI < 17 (IS reference); c consistent with WHO adult BMI < 16 (IS reference); d BAZ< -1 z-scores and ≥ - 2 z-scores (WHO reference); e BAZ < -2 z- scores and ≥ - 3 z-scores (WHO reference) ; f BAZ < -3 z-scores (WHO reference); g HAZ < -2 z-scores (WHO reference); h retinol<0.7 µmol/L; i retinol<1.05 and ≥ 0.7 µmol/L; j zinc<0.65 mg/L for children<10y, zinc<0.70 mg/L for fasting girls>10y and non fasting boys>10y, zinc <0.66 mg/L for non fasting girls >10y, zinc <0.74 μg/L for fasting boys>10y; k hb<11.5 g/dL (<12y), hb <12.0 g/dL (children<152y and girls>15y), hb<13.0 g/dL (boys>15y); l corrected ferritin <12 µg/L; m sTFR>8.3 mg/L

Anthropometric characteristics and micronutrient status of participants are shown in table 7, for all and disaggregated by age group. Less than 5% of the participants were stunted. Mean HAZ was significantly lower in adolescents than in children (P=0.02) and had a tendency to be lower in boys than in girls (P=0.08). Cumulate moderate and severe thinness measured with BAZ was much more prevalent, affecting almost 20% of participants without any statistical difference between boys and girls or age groups. Both severe thinness (BAZ <-3 z-scores) and grade 3 of thinness (equal to adult BMI <16 kg/cm²) were found in around 6% of children. Prevalence of moderate thinness (BAZ ≥ -3 z-scores and <-2 z-scores) was 13% while prevalence of grade 2 of thinness was 10% (equal to adult BMI between 17 and 16 kg/cm²). Mild thinness (BAZ ≥ -2 z-scores and <-1 z-scores) was 37% and slightly higher than thinness grade 1 whichwas 34% (equal to adult BMI between 16 and 18.5 kg/cm²). Only 3.0% of participants were overweight and two participants were obese (0.3%).

Fourteen percent of the participants were anemic. No significant difference was observed between gender groups but adolescents had a tendency to be more affected by anemia than children (P=0.07). Only 3 participants had severe anemia. Prevalence of low FER was 21% and prevalence of high TFR was 33%. Mean TFR was significantly higher in boys than in girls (P=0.02) as well as percentage of high TFR (P= 0.03) whereas there was no significant difference between children and adolescents. Prevalence of ID was found in approximately one third of participants, without any difference between age or gender groups. Prevalence of negative body iron content was 18%. Iron deficiency anemia was 11% when ID was defined by high TFR and/or low FER and 7% when ID was defined by negative body iron. While VAD was present in 3% of the participants, none had severe vitamin A deficiency. In contrast, approximately 40% of the participants had marginal VAD. Mean RBP was significantly lower in boys (P=0.01) than in girls and in children compared to adolescents (P=0.001). Moreover boys were significantly more affected by marginal VAD than girls (P=0.003; figure 2) and children were more affected than adolescents (P<0.0001). ZnD was highly prevalent at 26% of all participants and affected boys more than girls (P=0.02, figure 2). Median UIC was 137 µg/L with 7% of the participants having UIC <50 µg/L and 26% having UIC ≥50 µg/L and <100 µg/L. Only 1% had IUC <20 µg/L. Mean UIC was significantly higher in boys (P=0.04) and the prevalence of very high UIC (>300 µg/L) tended to be higher in boys (P=0.07, figure 2). 22% of participants had elevated UIC, 17% between 200 and 299 µg/L and 5% above 300 µg/L. The table 8 reports the public health significance of each nutritional disorder reported in the current paper according to international references 124-128 .

No association was found between micronutrient deficiency and thinness/stunting. Multiple micronutrient deficiency was not prevalent. No association has been found between iron deficiency, zinc deficiency, thinness, stunting, and socioeconomic characteristics (education level of the mother, socioeconomic status of the mother and the household head).

TABLE 8 PUBLIC HEALTH SIGNIFICANCE OF NUTRITIONAL DISORDERS IN CHILDREN FROM PRIMARY STATE SCHOOLS OF DAKAR

Indicators Prevalence Public health significance Vitamin A deficiency 3.0% mild Zinc deficiency 25.9% mild Iodine deficiency UI <50 µg/L 7% no disorders

median UIC 136.8 µg/L Anemia 14.4% mild Iron deficiency lowFer 21.4% yes highTfR 33.3% Thinness 50.6% (corresponding to BMI<18.5 high prevalence (serious kg/cm 2in adults) situation) Stunting prevalence 4.9% low prevalence

Discussion

The present study, which was carried out in a representative sample of school children attending state primary schools in Dakar, showed multiple nutritional problems. Prevalence of thinness, anemia and deficiencies of iron and zinc were high. In contrast, overweight and vitamin A deficiency were less prevalent. More than 50% of the children had evidence for inadequate (too low or too high) iodine intakes.

In the present study, both references from WHO and IS have been used because they are relatively new and under-used: in 376 previous studies on school-children, neither references were used 17 . The comparison between the 2 references shows only slight differences. Prevalence of grade 3 thinness as defined by Cole was similar to severe thinness defined by WHO, which is consistent with a recent study carried out in Seychelles 204 . Prevalence of grade 2 and 1 thinness (Cole) were slightly lower than moderate and mild thinness respectively as defined by WHO. As reported earlier, the WHO reference gives higher prevalence of especially mild thinness, regardless of age of the subjects. Many factors underline the high prevalence of thinness observed in the present study. For instance, infections or communicable diseases could have contributed to thinness. In our study, 12% of children had signs of inflammation as indicated by the elevated concentrations of acute phase proteins. Malaria, estimated to affect 8% of children, or diarrhea, are major causes of disease in Senegal. Other infections such as HIV/AIDS or tuberculosis are less prevalent in Senegal, with 0.7% of adults and 0.4% of people aged 15-24 y being HIV-positive 205 . and 0.2% of adult population is affected by tuberculosis 206 One might therefore assume that thinness is closely related to low dietary intake. It has been demonstrated that whereas malnutrition in infancy and young childhood is strongly related to stunting, thinness is an indicator of malnutrition in all age groups and suggest recent undernutrition 207 . The low prevalence of stunting (5%) would indicate that nutritional intake in the first years of life was probably adequate, with under nutrition only appearing later in childhood.

A study on food practices and nutrition in urban regions of Senegal identified several reasons for food insecurity and undernutrition inurban school-aged children 145 . These are mainly related to changing dietary habits. For instance, urban populations tend to decrease domestic foods in favor of street foods, with only one meal being home-cooked and consumed in the afternoon. For breakfast and dinner, families tend to increasingly buy snacks in the street. Worryingly, children attending urban state schools do not have lunch at home, while simultaneously not having access to canteens and thus school meals. During school days, they buy cheap foods from street sellers, which most often lack in proteins and micronutrients. This is highlighted by the finding that in the present study that found almost 6% of the participants were categorized as severe malnourished (BAZ ‹ -3 while only 1.9% of children under 5 y living in urban areas of Senegal have weight for height <-3 z-score 185 . Advocacy and more importantly actions for intervention improving the nutritional status of school children in Senegal are thus vital. Offering affordable and healthy meals in school canteens is but only one option available.

The prevalence of anemia (14%) still represents a public health problem 208 . This prevalence was less than half of the prevalence of anemia reported in school children in Dakar 18 years ago (35%) 191 . Another study carried out in Dakar in 2003 on a small sample of seven years old children indicated a prevalence of anemia of 39% 209 . These combined data suggest that anemia has decreased among Senegalese school children in Dakar over the last decade. Compared to other African countries, the prevalence of anemia found in the present study was much lower than the prevalence of ~40% recently found in 2 studies among school children in Burkina Faso and Cote d’Ivoire 210 . In contrast, a recent study carried out in 6-16 y old school children in rural Kenitra, Morocco found a similar anemia prevalence of 12% 211 .

Prevalence of ID defined either by low ferritin and/or by high transferring receptor concentrations was high (39%). However, in this study more participants had elevated TFR (33%) than low ferritin (21%) or negative body iron (18%). Both indicators measure different stages of iron deficiency, with low ferritin indicating the depletion of iron stores and elevated TFR related to the iron-deficient erythropoiesis, indicating a later stage of iron deficiency 210 . To our knowledge, it is the first time that ID in a population was more related to abnormal TFR values than to low ferritin. However, a study carried out in 5-15 y old children in Cote d'Ivoire recommends the use of higher cut-offs for TFR i.e. 9.4 mg/L for African populations (versus 8.3 mg/L used in our study) to improve its performance in defining iron status in children. However, the authors concluded that TFR had only a modest sensitivity and specificity in identifying iron deficiency, regardless of the diagnostic cutoffs chosen 210 . When applying this cut-off, the prevalence of high TFR decreased from 33 to 20 %, which was very close to the 18% of children with a negative total body iron. This study suggests that a cut-off of 9.4 mg/L for TFR might be better in these settings. Even with this higher cut-off, the prevalence of ID defined by either abnormal ferritin or TFR remained high at 29%.

ID prevalence was more than double the prevalence of anemia demonstrating that anemia prevalence rates cannot be used as a proxy indicator for ID 198 To allow correction of ferritin concentrations in the presence of inflammation, iron status indicators have to be included in epidemiologic studies combined with indicators of inflammation. In addition, the prevalence of iron deficiency anemia (IDA) was low and only represented half of the anemia prevalence, indicating that other factors contributed to anemia. Malaria infection and hemoglobinopathies

52 such as sickle cell are likely contributors with prevalences of 8% and 10% respectively 212,213 The prevalence of vitamin A deficiency, which can contribute to anemia, was low. However, other micronutrients such as vitamins B12 and folic acid were not measured. These deficiencies could have played a role in the etiology of the non-ID anemia found in the present study 214 .

In Burkina Faso, 40% of school children aged 7-14 y in Ouagadougou are vitamin A deficient 215,216 , whereas 59% of children aged 6-9 y are vitamin A deficient and 8% have severe VAD in rural Northern Ethiopia 217 . Both studies used retinol concentrations as an indicator while the current study uses RBP concentrations. However, at low plasma retinol concentrations, RBP concentrations is a less sensitive indicator of vitamin A status, as more unbound RBP appears in the circulation 104,218 . Using RBP concentrations may therefore have underestimated the prevalence of vitamin A deficiency in the current study. Mean RBP concentration was significantly lower and prevalence of marginal vitamin A status higher in boys and in children. Although the study notes the differences in vitamin A deficiencies between boys and girls, with higher risks for boys, it does not provide a clear answer for this variance. However, similar sex differences for other micronutrients such as iron and zinc have been reported before 219 .

FIGURE 2 GENDER -RELATED DIFFERENCES IN PREVALENCE OF ABNORMAL STATUS Zinc deficiency (ZnD) was highly prevalent with almost 1/4of all school children affected representing a significant higher rate than the cut-off of 20% indicating a public health problem 126 . Furthermore, in consensus with similar studies, boys were significantly more affected than girls. The higher requirements of zinc for boys generally suggest that boys are more sensitive to ZnD 220-223 Indeed, boys have higher proportion of muscles per kilogram body weight, which contains a higher content of zinc than fat, and the growth rate of boys is higher than girls 203 .

According to both the median UIC (136.8 µg/L) and proportion of children with UI<50 µg/L (7%), iodine deficiency is not a major public health issue in school children in Dakar. Iodine nutrition in Dakar is supposed to be good, as Senegal introduced iodized salt in 1995. According to a national survey in 2006, 70.5% of households consume adequate iodized salt (>15 ppm) 185 . Moreover, Dakar’s geographical location near the sea ensures adequate intake from seafood 185 .. Nonetheless, 33% of children had insufficient iodine intake (UIC below 100 µg/L), and although this is lower than WHO estimates for Africa in general (41% of children 6-12 years having UIC‹100 µg/L 5 it is still a considerable percentage of children. Worrying, only 45% of the

53 children had adequate iodine intake (UIC between 100 and 199 µg/L), with 22% of the children being at risk for iodine-induced hyperthyroidism (UIC above 200 µg/L) and 5% of children risk adverse health consequences (UIC above 300 µg/L) like iodine-induced hyperthyroidism or autoimmune thyroid disease) 124 Two studies in non coastal countries Lesotho and Angola on children from rural schools reported respectively 94% and 78% of children with UIC<50 µg/L and only 0% and 2% children with UIC≥200 µg/L. As iodine deficient populations seem to be more sensitive to health consequences of excessive iodine intakes 5, urgent attention is needed for iodine nutrition in school children in Dakar. Apart from gender and age related differences, it was not possible to identify a specific vulnerable socioeconomic group of children, or a group of children more affected by malnutrition. Nutritional issues seem to randomly affect the whole population of the study.

Not including non-school attendees is a limitation of the study, as they might be the most vulnerable to malnutrition. However, one of the objectives of the study was to provide information to the education ministry in order to illustrate nutritional disorders in urban school attendees. The goal was to reorient the strategy of school-feeding programs towards broader coverage, as only rural state schools received school meals. The entire population of urban school-aged children and adolescents should be taken into account by nutrition policy.

Conclusion

To conclude, many school-aged children in urban Senegal have a poor nutritional status, as the high prevalence of iron and zinc deficiency illustrate. Furthermore, iodine intake as indicated by urinary iodine concentrations was either too low or too high in over half of the children. Although the low rate of stunting in the population suggests adequate nutrition during the first years of life, the prevalence of thinness going up to almost 20% remains alarming. The study highlighted that the transition from home meals in the preschool period to self-catering at school is most likely the basis for these multiplex of nutritional problems. The study therefore stresses the need for nutritional interventions to improve dietary quality and quantity of school children in Senegal.

2.2. Nutrient intake is insufficient among African urban school-aged children and adolescents: results from two 24-hours recall in primary state schools in Dakar, Senegal (Manuscript, drafted)

To be submitted Marion Fiorentino a, Edwige Landais a, Alicia Carriquiry d, Guillaume Bastard b, Frank T. Wieringa a, Jacques Berger a

Introduction

Malnutrition is estimated to be highly prevalent in African school-aged children, with 20% of stunting, 30% of thinness, anemia, iron, iodine or vitamin A deficiency, 50% of zinc deficiency and 10% of overweight 17 . Due to growth, cognition and educational achievements, nutrient requirements of school-aged children (5 – 9 years (y) ) are high 17 . Because of late school enrollment often occurring in low and middle income countries (LMIC) primary school age also includes early adolescence, a period when skeleton and brain growth spurts and sexual maturation dramatically increase nutrient requirements 17 . Early marriage, still common in the developing world 26 , makes adolescent girls particularly vulnerable to undernutrition in case of childbearing 108,117 . Thus, nutrient deficiencies may be especially harmful for cognitive and physical development during school age and early adolescence (10 - 14 y). Although physiological status and health conditions like menarche or parasitosis have an influence on nutritional status, nutritional status is mainly related to nutrient intake. Thus, dietary intake data are crucial to understand the role of the diet in nutritional deficiencies, and eventually to design effective nutritional interventions.

In LMIC, diet of school-aged children and adolescents is usually poor in fruits, vegetables and animal products, leading to inadequate protein and micronutrient intake 151 . Moreover, urban new lifestyles increase sedentary activities and consumption of high fat and sugar snacks and beverages that are often energy dense but micronutrient poor 151 . Although living in urban areas was known to improve dietary diversity and therefore nutritional status 224 , some evidence showed that in poor urban settings micronutrient deficiencies are still prevalent 215 and can coexist with increasing overweight. Indeed, in African urban areas, school-aged children spend more time out-of-home away from their caretaker supervision and consume more street food compared to their younger counterparts 145 . Thus, changes in food patterns due to the transition from preschool to school age can affect nutrient intakes especially when no school feeding programs are available. Consequently, populations of school-aged children and adolescents living in African urban areas may be at risk for double burden of malnutrition associated with inadequate dietary intake.

Recent data on nutrient intake adequacy and its implication for nutrition in primary school children living in African urban settings is scarce. No data on dietary intake of school-aged children and adolescents from Senegal were available 151 . The authors conducted a cross- sectional study investigating anthropometry, micronutrient status and dietary intake among a representative sample of 600 children attending primary state schools in Dakar, Senegal. Results from anthropometric and biochemical data were published elsewhere 225 . Chronic malnutrition was low but acute malnutrition was prevalent, with less than 5% of the children

55 being stunted and 18% of children having BMI-for-age <−2 z -scores. Overweight and obesity were not prevalent in that population. Micronutrient status was poor with 36% of children suffering from marginal vitamin A status, 26% being zinc deficient and 39% being iron deficient. The objectives of the present article were i) to estimate nutrient intake ii) to report prevalence of insufficient/excessive nutrient intake – for all and by age and gender groups iii) to evaluate inadequate nutrient intake as risk factors for micronutrient deficiencies measured on blood samples.

Participants and methods

Study design and participants

The study was a representative cross-sectional survey targeting children registered in primary state schools in the area of Dakar. 600 school children were randomly selected through a two- stage random cluster sampling of children attending primary state schools in the Dakar area (30 schools × 20 children). Details of study design and sampling are provided elsewhere 225 . Adolescents were defined as equal or above 10 y 14 .

Data collection

The study was conducted over 30 school-days in February-March 2010 by trained interviewers. Two quantitative 24-hours recalls (24-hr) were conducted, at school, with three school days between repeated recalls. Children were asked to provide details on all foods and beverages consumed during the previous day. Rice and cereal gruels were quantified by weighting salted replicas to the nearest gram, using an electronic scale Terraillon®. The other foods were quantified using pictures of different portion size 226 , household measures, standard units or purchase price. During the afternoon following the 24-hr, mothers or caretakers were asked at home about recipes of each home-cooked food or beverage mentioned by the child during the 24-hr. The amount of each ingredient was estimated in grams, household measures or price. Similarly, recipes of foods (snacks and beverages) sold inside or outside schools were recorded by interviewing vendors. During the second 24-hr, anthropometric measurements (height and weight), blood and urine samples were collected to measure vitamin A, zinc, iron and iodine status.

Data management and analysis

Data entry, including quality checks and validation by double entry of questionnaires, was performed with epidata entry version 3.1 (epidata Association, Odense, Denmark). A food composition table compiled from different food composition tables 227-230 was specifically developed for the study. Data management was performed using epidata Analysis version 2.2 (epidata Association, Odense, Denmark). Energy and nutrient intakes from repeated 24-hr were adjusted for intra individual variability to obtain estimated usual intakes, using PC-Side (PC Software for Intake Distribution Estimation, ISU, 1997, Chicago) 231 . Estimated energy requirement (EER) were calculated using Schofield equations taking into account height, weight and age, Physical Activity Level (PAL) for active and low active children 232. Cut-offs used to estimate insufficient and excessive intakes are summarized in the table 9 233-235 . Prevalence of insufficient and excessive nutrient intake was estimated using PC-Side. Prevalence of

56 insufficient nutrient intake was secondly calculated based on nutrient intake not adjusted for day-to-day variability and presented in the supplementary material 2. Crude odds ratios (OR) and odds ratios adjusted for age group and gender were calculated to measure the association between insufficient nutrient intake and micronutrient deficiencies.

Ethics

TABLE 9 RECOMMENDATIONS OF DAILY INTAKE OF MACRONUTRIENTS AND MICRONUTRIENTS FOR POPULATIONS OF CHILDREN 4-18 Y

Vitamin Folic Vitamin Zinc Iron Calcium Carbohydrate Protein Lipid SFA MUFA PUFA Fiber A acid C

Units μg μg mg mg mg mg g %EI g/kg %EI %EI %EI %EI %EI g

Insufficient intake

Cut-off EAR EAR EAR EAR EAR AI EAR AMDR EAR AMDR AMDR - - - AI

4-8y 275 160 22 4 4.1 800 100 0.76 25

9-13y boys 445 250 39 7 5.9 1300 100 0.76 31 45 10 25 - - 5* 9-13y girls 420 250 39 7 5.7 1300 100 0.76 26

14-18y boys 630 330 63 8.5 7.7 1300 100 0.73 38

14-18y girls 485 330 56 7.3 7.9 1300 100 0.71 26

Excessive intake Cut-off UL UL UL UL UL UL - AMDR - AMDR AMDR - - - -

4-8y 900 400 650 12 40 2500 within 9-13y 1700 600 1200 23 40 2500 limits - 65 - 30 35 10* of 15* -

total fat* 14-18y 2800 800 1800 34 45 2500

These cut-offs are recommended in the essential guide to nutrient requirements of the Institute of Medicine (IOM) 233 ] except * 234 SFA: Saturated Fatty Acid; MUFA: Monounsatured Fatty Acid; PUFA : Polyunsaturated Fatty Acid; %EI : % of energy intake; EAR : Estimated Average Requirement; AI : Adequate Intake; AMDR : Acceptable Macronutrient Distribution Ranges; UL : Upper Limit

NB : In the present study, 167 (31%) participants were <9 y, 160 (29%) participants were boys aged from 9 to 13 years, 182 (33%) participants were girls aged from 9 to 1 3 y, 19 (3%) participants were boys ≥14 y, 17 (3%) participants were girls ≥14 y

Results

Dietary intake was obtained for 599 participants aged 5 to 17 y. Fifty four children identified as misreporters were excluded, leading to a final sample of 545 children. 52.6% were girls and 45.1% were <10 y, resulting in 116 boys<10 y, 130 girls<10 y, 142 adolescent boys and 157 adolescent girls. Most of children had insufficient energy intake, 7% had insufficient protein intake and 88% had insufficient fiber intake. None of them had carbohydrates intake below recommendation. Mean contribution to energy from carbohydrates, proteins, lipids, PUFAs and SFAs were within the AMDRs but the contribution of proteins and carbohydrates to energy was insufficient in 31% and 5% of children, respectively. No children had excessive contribution of protein and carbohydrates to energy, but respectively 32%, 21% and 14% of children had excessive contribution of lipids, SFAs, and PUFAs (Table 10). Iron and vitamin C intakes were insufficient for half of the children (Table 11) and zinc and vitamin A intakes were insufficient for more than two thirds. Calcium and folic acid intakes were insufficient in all children. No micronutrient was consumed above their respective Upper Limits. Prevalence of inadequate nutrient intake and contribution of macronutrients to energy were different between age groups. Prevalence of insufficient contribution of carbohydrates to energy was higher in adolescents compared to children <10 y in both genders (Figure 3). Protein intake was lower in adolescents compared to children <10 y, among boys (1.17±0.06 g/kg vs. 1.49±0.08 g/kg, P<0.001) and among girls (1.05±0.05 g/kg vs. 1.41±0.08 g/kg P<0.001). Insufficient protein intake was more prevalent among adolescent boys compared to boys <10 y (8±5% vs. 1±2%, P<0.05). PUFAs’ contribution to energy was higher in adolescents compared to children <10 y, in boys (13.3% vs. 12.3%, P<0.001) and in girls (12.6% vs. 11.6%, P<0.001), and so was prevalence of excessive contribution of lipid and PUFA to energy (Figure 3). Prevalence of excessive contribution of SFAs to energy was higher in girls <10y compared to adolescent girls. Insufficient zinc intake was more prevalent in adolescents compared to children <10 y in both genders (Figure 4). Among girls, prevalence of insufficient vitamin C intake was higher in adolescents compared to girls <10y (Figure 4). Among boys, prevalence of insufficient vitamin A intake was higher in adolescents compared to boys <10y.

Differences in prevalence of inadequate intake according to gender were also observed. Intakes were higher in adolescent boys compared to adolescent girls in energy (1507±45 kcal vs. 1399±38 kcal, P<0.001), in carbohydrates (193±6 g vs. 183±6 g, P <0.05), in proteins (39±2g vs. 36±1g, 1.17±0.06g/kg vs. 1.05±0.05g/kg, P<0.01 for both), in fibers (20±1g vs. 18±1g, P <0.001), in lipids (58±2 g vs. 53±2 g, P<0.001), in SFAs (14±1g vs. 13±1 g, P<0.05), and in PUFAs (23±1g vs. 20±1g, P<0.001). Proportion of daily energy coming from PUFAs was higher in boys compared to girls, among children <10y (12.3±0.4 % vs. 11.6±0.3%, P<0.001) and among adolescents (13.3±0.4% vs. 12.6±0.5%, P<0.05). Excessive proportion of PUFAs in total energy was more prevalent in boys<10y compared to girls<10y (Figure 3). Mean vitamin C intake was lower in boys<10y compared to girls <10y (24±2 mg vs. 34±2 mg, P<0.001). Mean folic acid and zinc intake were also lower in boys <10y compared to girls <10y (40±2 µg vs. 47±3 µg, P<0.001; 4.6±0.3 mg vs. 5.1±0.3 mg, P<0.05). Mean folic acid intake was higher in adolescent boys compared to adolescent girls (60±3 µg vs. 51±3 µg, P<0.001). Prevalence of insufficient zinc intake was significantly higher in adolescent girls compared to adolescent boys (Figure 4).

The risk for iron deficiency was higher when intake of iron and protein were under EAR and when intake were high in energy from lipids and carbohydrates (Table 12). Zinc deficiency risk was higher when zinc intake was insufficient, when contribution of protein to energy intake was insufficient,

59 when contribution of lipid and SFA was excessive and it was lower when fiber intake was below EAR. Risk for vitamin A marginal status was higher when iron intake, vitamin C intake and contribution of protein to energy intake was insufficient, and lower when energy intake from lipid was high.

TABLE 10 MEAN ENERGY AND MACRONUTRIENT DAILY INTAKE (ADJUSTED FOR WITHIN -PERSON VARIABILITY ) AND PREVALENCE OF INSUFFICIENT AND EXCESSIVE MACRONUTRIENT DAILY INTAKE

mean intake/ proportion of ± inadequate intake Energy (kcal) 1365 23 Energy/EER 1 0.65 0.01 Children with energy intake below EE1 (%) 99 1 Energy/EER 2 0.75 0.01 Children with energy intake below EE2 (%) 93 2 Protein (g) 36 1 Protein/body weight (g/Kg) 1.25 0.04 Children with protein intake /body weight below EAR (%) 7% 2% % Energy from protein 11% 0% Children with % energy intake from protein below AMDR (%) 31% 4% Children with % energy intake from protein above AMDR (%) 0% 0% Carbohydrates (g) 176 3 Children with fiber intake below EAR (%) 0% 1% % Energy from carbohydrates 53% 0.4% Children with % energy intake from carbohydrates below AMDR (%) 5% 1.8% Children with % energy intake from carbohydrates above AMDR (%) 0% 0.3% Dietary fiber (g) 19 1 Children with fiber intake below AI (%) 88% 3% Lipid (g) 52 1 % Energy from lipid 33% 0.3% Children with % energy intake from lipid below AMDR (%) 1% 0.8% Children with % energy intake from lipid above AMDR (%) 32% 3.9% SFA (g) 13 0 % Energy from SFA 8% 0% Children with % energy intake from SFA above 10% 21% 3% MUFA (g) 15 0 % Energy from MUFA 9% 0.1% PUFA (g) 19 0 % Energy from PUFA 12% 0.2% Children with % energy intake from PUFA below 5% (%) 0% 0.0% Children with % energy intake from PUFA above 15% 14% 2.9% AMDR: Acceptable Macronutrient Distribution Ranges; AI : Adequate Intake; EAR : Estimated Average Requirements; EER1 : Estimated Energy Requirement for active children ; EER2 : Estimated Energy Requirement for low active children; SFA : Saturated Fatty Acids; MUFA : Mono-unsaturated Fatty Acids; PUFA : Poly-unsaturated Fatty Acids

TABLE 11 MEAN MICRONUTRIENT DAILY INTAKE (ADJUSTED FOR WITHIN -PERSON VARIABILITY ) AND PREVALENCE OF INSUFFICIENT AND EXCESSIVE MICRONUTRIENT DAILY INTAKE

mean intake/ proportion of ± inadequate intake Iron (mg) 5.6 0.1 Children with iron intake below EAR (%) 46% 4% Children with iron intake above UL (%) 0% 0%

Zinc (mg) 5.3 0.2 Children with zinc intake below EAR (%) 69% 4% Children with zinc intake above UL (%) 1% 1%

Calcium (mg) 268 8 Children with calcium intake below AI (%) 100% 0% Children with calcium intake above UL (%) 0% 0%

Vitamin C (mg) 38 2 Children with vitamin C intake below EAR (%) 53% 4% Children with vitamin C intake above UL (%) 0% 0%

Folic acid (µg) 50 2 Children with folic acid intake below EAR (%) 100% 0% Children with folic acid intake above UL (%) 0% 0%

Vitamin A (µg) 309 17 Children with vitamin A intake below EAR (%) 79% 3% Children with vitamin A intake above UL (%) 0% 0%

AI: Adequate Intake; EAR: Estimated Average Requirement; UL: Upper Limit

TABLE 12 CRUDE AND AGE -GENDER ADJUSTED ODDS RATIOS (OR ) AND CONFIDENCE INTERVAL (CI ) FOR INSUFFICIENT NUTRIENT INTAKE ASSOCIATED WITH MICRONUTRIENT DEFICIENCIES

Iron deficiency Marginal vitamin A status Zinc deficiency

OR 95% CI OR 95% CI OR 95% CI

Iron intake below EAR Crude 1.32 [ 0.93 ; 1.88 ] 1.84 [ 1.26 ; 2.68 ] 1.05 [ 0.74 ; 1.48 ] Adjusted * 1.47 [ 1.02 ; 2.13 ] 1.80 [ 1.22 ; 2.65 ] 1.35 [ 0.93 ; 1.96 ]

Zinc intake below EAR Crude 1.24 [ 0.83 ; 1.84 ] 1.53 [ 1 ; 2.33 ] 1.24 [ 0.84 ; 1.82 ] Adjusted * 1.46 [ 0.95 ; 2.26 ] 1.27 [ 0.8 ; 2.02 ] 1.76 [ 1.13 ; 2.72 ]

Vitamin C intake below EAR Crude 0.74 [ 0.52 ; 1.06 ] 1.93 [ 1.32 ; 2.83 ] 1.24 [ 0.87 ; 1.78 ] Adjusted * 0.83 [ 0.55 ; 1.24 ] 1.80 [ 1.21 ; 2.68 ] 1.00 [ 0.53 ; 1.87 ]

Fiber intake below AI Crude 0.68 [ 0.39 ; 1.16 ] 0.88 [ 0.51 ; 1.5 ] 0.09 [ 0.05 ; 0.14 ] Adjusted * 0.59 [ 0.32 ; 1.11 ] 1.13 [ 0.64 ; 1.98 ] 0.39 [ 0.17 ; 0.88 ]

Vitamin A intake below EAR Crude 0.87 [ 0.56 ; 1.35 ] 1.73 [ 1.07 ; 2.81 ] 1.12 [ 0.73 ; 1.72 ] Adjusted * 0.72 [ 0.43 ; 1.18 ] 1.16 [ 0.68 ; 1.96 ] 1.20 [ 0.73 ; 1.98 ]

Protein per kg of body weight below EAR Crude 2.20 [ 1.17 ; 4.13 ] 0.30 [ 0.13 ; 0.67 ] 0.36 [ 0.12 ; 1.07 ] Adjusted * 2.22 [ 1.14 ; 4.33 ] 0.44 [ 0.19 ; 1.03 ] 0.61 [ 0.19 ; 2.02 ]

% of energy coming from protein below AMDR Crude 0.25 [ 0.16 ; 0.39 ] 4.58 [ 3.03 ; 6.92 ] 2.91 [ 1.98 ; 4.29 ] Adjusted * 0.34 [ 0.22 ; 0.54 ] 3.28 [ 2.07 ; 5.19 ] 3.00 [ 1.91 ; 4.69 ]

% of energy coming from carbohydrates above AMDR Crude 0.17 [ 0.08 ; 0.36 ] 2.10 [ 0.84 ; 5.3 ] 4.26 [ 2.47 ; 7.35 ] Adjusted * 0.21 [ 0.1 ; 0.45 ] 1.87 [ 0.69 ; 5.05 ] 1.36 [ 0.66 ; 2.81 ]

% of energy coming from lipid above AMDR Crude 2.04 [ 1.41 ; 2.95 ] 0.48 [ 0.31 ; 0.75 ] 1.93 [ 1.33 ; 2.8 ] Adjusted * 2.02 [ 1.39 ; 2.93 ] 0.47 [ 0.29 ; 0.74 ] 1.67 [ 1.09 ; 2.56 ]

% of energy coming from SFA above AMDR Crude 1.44 [ 0.91 ; 2.29 ] 0.53 [ 0.31 ; 0.92 ] 3.31 [ 2.04 ; 5.38 ] Adjusted * 1.43 [ 0.9 ; 2.28 ] 0.57 [ 0.32 ; 1 ] 2.89 [ 1.7 ; 4.91 ]

% of energy coming from PUFA above AMDR Crude 2.90 [ 1.79 ; 4.68 ] 0.37 [ 0.19 ; 0.69 ] 0.93 [ 0.58 ; 1.49 ] Adjusted * 3.02 [ 1.85 ; 4.91 ] 0.51 [ 0.26 ; 1 ] 0.99 [ 0.59 ; 1.66 ]

*Adjusted for age group and gender AI : Adequate Intake; AMDR : Acceptable Macronutrient Distribution Ranges; EAR : Estimated Average Requirement; PUFA: Polyunsaturated Fatty Acid; SFA: Saturated Fatty Acid

1% # 5% Insufficient protein intake /body weight 8% # 5%

34% 26% % EI from protein below AMDR (10%) 38% 32% # 3% 0% oo % EI from carbohydrates below AMDR (45%) 10% # 7% oo

91% * 80% Insufficient fiber intake * ooo 94% 94% ooo 0% 0% o % EI from lipid below AMDR (25%) 2% 5% o

26% # 20% % EI from lipid above AMDR (35%) o 40% # 31% o 23% 30% oo % EI from SFA above AMDR (10%) 16% 14% oo

9% * # 2% % EI from PUF above AMDR (15%) * ooo 22% # 19% ooo

Boys <10y Girls <10y Boys ≥10y Girls ≥10y

FIGURE 3 PREVALENCE OF INSUFFICIENT MACRONUTRIENT INTAKE ACCORDING TO AGE AND GENDER GROUP

%EI: percent of energy intake; AMDR : Acceptable Macronutrient Distribution Range P-value for χ 2-test between boys <10y and girls <10y : ***: P-value<0.001; **: P-value<0.01; * P-value <0.05 P-value for χ 2-test between boys ≥10y and girls ≥10y : +++: P -value<0.001; ++: P-value<0.01; + P-value <0.05 P-value for χ 2-test between boys <10y and boys ≥10y : ###: P -value<0.001; ##: P-value<0.01; # P-value <0.05 P-value for χ 2-test between girls <10y and girls ≥10y: ooo: P -value<0.001; oo: P-value<0.01; o P-value <0.05

FIGURE 4 PREVALENCE OF INSUFFICIENT MICRONUTRIENT INTAKE ACCORDING TO AGE AND GENDER GROUP

P-value for χ2 -test between boys <10y and girls <10y : ***: P-value<0.001; **: P-value<0.01; * P-value <0.05 P-value for χ2 -test between boys ≥10y and girls ≥10y : +++: P -value<0.001; ++: P-value<0.01; + P-value <0.05 P-value for χ2 -test between boys <10y and boys ≥10y : ###: P -value<0.001; ##: P-value<0.01; # P-value <0.05 P-value for χ2 -test between girls <10y and girls ≥10y: ooo: P -value<0.001; oo: P-value<0.01; o P-value <0.05

FIGURE 5 UNADJUSTED AND ADJUSTED PREVALENCE OF INSUFFICIENT AND EXCESSIVE INTAKE

%EI : % of energy intake; AI : Adequate Intake; AMDR : Acceptable Macronutrient Distribution Ranges; EAR : Estimated Average Requirement; PUFA : Polyunsaturated Fatty Acid; SFA: Saturated Fatty Acid; UL : Upper Limit

Discussion

Energy, macronutrients and fiber intake

Energy intake was insufficient in most children and the distribution of macronutrient was unbalanced with excessive lipid intake, especially from SFA, and insufficient protein intake for one third of the children. Few data on energy and macronutrient intake are available for African children but a study carried out in Cameroonian urban adolescents also indicated unbalanced diets with excessive carbohydrates and insufficient protein contributions to energy 236 . In the current study, lipid and SFA excessively contributed to energy intake and it contrasts with a study in Ghanaian rural school children where energy intake from carbohydrates and lipids were respectively higher and lower compared to the present study 237 . Most children had fiber intake below the Adequate Intake (AI). Even if fibers are not essential nutrients of which insufficient intakes lead to biochemical or clinical deficiency symptoms AI has been established to define fiber intake providing the greatest protection against coronary heart disease 238 . Moreover, dietary fiber is considered as protective against inflammatory bowel disease and colorectal cancer 239,240 , of which excessive fat consumption, especially SFA, is suspected to be a risk factor 241 . Insufficient protein intake and contribution of protein to energy intake were respectively risk factors for iron and zinc deficiency. This is not surprising as food rich in zinc and iron are usually also rich in protein (meat, shellfish, legumes), thus diets lacking of these products may lead to concomitant protein, iron and zinc poor status 242 . In the present study, excessive energy coming from lipids and from SFA were also risk factors for zinc deficiency, which is notable as some research suggests that the type of dietary fat influences the effects of zinc deficiency on fatty acid status and especially lipid concentrations in the liver 243 . We also observed that excessive lipid and PUFA intake was positively associated with iron deficiency. This could compensate the PUFA status decline induced by iron deficiency which is suggested in other research 244 . Indeed, Smuts et al assumed a possible disruption of PUFA metabolism in iron deficient children when they observed higher SFA status and lower PUFA status among iron deficient school children compared to children with normal iron status. Not surprisingly, in our study insufficient fiber intake was a protective factor for zinc deficiency, as dietary fiber is known to impair zinc absorption 238 . Even if energy intake increased with age, insufficient fiber, protein and micronutrient intake and inadequate contribution of macronutrients to total energy intake were higher among adolescent compared to children<10y, which suggest poorer nutrition in adolescents, where mean height for age was lower in the present study 225 . A research conducted in Cameroonian urban adolescents also reported a high prevalence of insufficient contribution of proteins to energy intake 236 . These findings suggest that in African urban areas, the increase of macronutrient intake with age does not compensate drastically increased requirements at adolescence. Adolescent girls had the lowest ratio of energy intake by estimated energy requirement. Macronutrient intake and nutritional status of adolescent girls is particularly worrisome, as thinness of adolescent mothers is associated with low birth weight of their babies 17 .

Iron intake

In the present study, iron intake was poor similarly to what was reported in Ghanaian 237 and Kenyan school children 245 . Because of high consumption of tea and low meat intake among African urban children 246 the absorption of iron in the present population is probably low 85 and therefore prevalence of insufficient iron intake based on EARs may be underestimated. Although peers reported that iron intake does not predict well serum ferritin concentrations 247 , in our study, insufficient iron intake was a risk to be iron deficient. Consequently, iron deficiency which affects 39% of school children of Dakar is likely to result from insufficient iron intake. By causing anemia and reducing physical and cognitive capacities, iron deficiency may impair school attendance and performance in the studied population 17 . Iron requirements are very high in adolescents especially in girls where menarche usually follows the growth spurt 85 . Insufficient iron intake was a risk factor for marginal vitamin A status. Some research suggested that iron deficiency may induce vitamin A deficiency by inhibiting mobilization of vitamin A store in rats 248 but the impact of iron deficiency on vitamin A metabolism remains unclear.

Zinc intake

Insufficient zinc intake was highly prevalent in children, as reported in other studies in school children and adolescents from urban Cameroon 236 , rural Ghana 237,249 and Kenya 237 . As reported in the literature 250 , in the present study, insufficient zinc intake was a good predictor of serum zinc concentrations and therefore a good predictor of risk of zinc deficiency. Poor zinc intake and zinc deficiency may disturb growth, immune system and appetite and increase morbidity from diarrhea and acute respiratory infections 85 . Albeit higher zinc intake, insufficient zinc intake was more prevalent in the adolescent groups compared to children < 10 y , probably due to much higher requirements associated with growth spurt 85 , where zinc has a central role 85 .

Calcium intake

In the present study, mean dietary calcium was below recommendations 251,252 ,238 and all children had insufficient calcium intake. Very high proportions of insufficient calcium intake were also reported in Cameroonian and Moroccan urban adolescents 236,253 , and in school children in Ghana 237 . Dietary calcium has been shown to be correlated to serum calcium levels in African school children and adolescents 254 and low dietary calcium is associated with hypocalcaemia 255 and decreased bone mineral density 254 . So it is assumed that school children of Dakar are at high risk for calcium deficiency. Although rickets mostly affect infants, it can also occur during late childhood and adolescence if diet is poor in calcium or in vitamin D and/or high in phytates 256 . Therefore, although cut-offs used in the present study may be considered as too high, in other research on osteomalacia, mean levels of calcium intake in school children and adolescents suffering from rickets ranged from 150 to 300 mg, which is similar to the present study 256-258 . Hence the population in the present study is at high risk for calcium deficiency, and therefore for rickets and osteomalacia 259 . The insufficient intake of calcium in the present study may be surprising, as consumption of dairy products is culturally part of the Senegalese diet 260 . However their high price in urban areas reduces their accessibility.

Vitamin C intake

Half of children had insufficient vitamin C intake. Elevated proportion of poor vitamin C intake was also observed in Kenyan and Ghanaian school children 237,245 as well as among Cameroonian and Nigerian adolescents 236,261 . As vitamin C intake was shown to be a good predictor of vitamin C serum values 250,262 , school children from Dakar are likely to be at risk for vitamin C deficiency. A diet low in fruits rich in both vitamin A and vitamin C could explain why insufficient vitamin C intake was associated with poor vitamin A status in the present study 263 . Vitamin C deficiency results in skeletal pain, increased susceptibility to infections, impaired wound healing at a moderate stage, and scurvy at a severe stage 85 . Vitamin C has antioxidant properties 264-266 and is known to improve immune response and to reduce the incidence of infectious diseases in children from developing countries 267 . It also stimulates iron absorption and metabolism, especially for non-heme iron 268,269 . High cost and instability during storage are the major obstacles to use vitamin C in nutritional intervention programs 270,271 but distributing fruits through school feeding programs in African urban areas may improve vitamin C, iron, and vitamin A status, especially in vegetables based diets 237 .

Folic acid intake

In the present study, all children had folic acid insufficient intake and mean folic acid intake was lower than what was reported in rural Kenyan and Ghanaian school children 237,245 although diets in these populations were already considered as folic-acid deficient. No recent data on folic acid status in Senegalese populations was available but a high risk of folic acid deficiency is assumed in this population as folic acid intake predicts well serum folic acid concentrations 250,272 . Insufficient folic acid intake and folic acid deficiency may result in leucopenia, intestinal malabsorption, blood coagulation impairment, increased sensibility to infections 85 , and macrocytic anemia, which is the second most common type of nutritional anemia, after iron deficiency 44 . Among girls from the present sample, folic acid intake did not increase at adolescence, oppositely to most of other micronutrients, and is lower than in boys. Similarly, insufficient folic acid intake was found in more than three quarters of urban adolescents in studies in Cameroon and Morocco 236,253 . Thus, low folic acid intake in adolescent girls from African urban areas is worrisome, due to its serious adverse consequences during pregnancy on the development of the foetus 85 .

Vitamin A intake

Children from the present sample had poor vitamin A intake. Vitamin A intake is highly dependent on seasons 273 . In Senegal, mangos are only available from April to August, while the present study was conducted in February-March. A high proportion of children with insufficient vitamin A intake was also reported outside the mango season in a study conducted in Ghana 237 . Only 3% of children in the current study were vitamin A deficient, which is considered as mild public health issue and negligible compared to other urban school children from West Africa 215 . However, both prevalence of marginal vitamin A status 225 and insufficient vitamin A intake suggested a risk for vitamin A deficiency, which is involved in morbidity linked to diarrhea and measles, in growth retardation 85 and in a large part of blindness or severe visual impairment 274 . No significant difference between boys and girls in the proportion of insufficient vitamin A intake was observed but boys were more affected by marginal vitamin A status than girls 225 ,

68 which is consistent with previous research 275 . As growth rates under 10 y are higher in boys than in girls 85 , the higher requirement of vitamin A in boys may not be adequately taken into account in EARs before adolescence. Risk for vitamin A poor status is probably a condition appearing at school age, as more than 90% of Senegalese children under 5 y receive yearly vitamin A supplementation 276 . Percentage EI coming from lipid above AMDR was a protective factor for marginal vitamin A status. Indeed, an increased intake of dietary fat is likely to improve the vitamin A absorption 85 .

Dietary and environment determinants

African urban school children and adolescents are supposed to be more at risk of overweight 17 151 and to consume more meat, vegetables, cereals, milk products than their rural counterparts 224 . However, in our study, children had insufficient micronutrients and energy intake, especially from protein, and acute malnutrition and micronutrient deficiencies were prevalent, while no overweight and low chronic malnutrition were observed, which suggests adequate care and diet in early childhood and degradation of diet and nutritional status when reaching school age 225 . Thus, nutrient intake data as well as biochemical and anthropometric data from this study indicated recent energy, protein and micronutrients deprivation in children and adolescents from primary state schools of Dakar, most likely due to deficient and inadequate diet, lacking of quantity and diversity with minimum consumptions of animal foods and fruits and vegetables. Consequently, fat and SFA excessively contributes to the total dietary intake, partly because of consumption of processed food and street food rich in low-quality lipids or sugar and poor in protein 151 . Indeed, it is known that urban lifestyle added to the food price crisis and global crisis in the 2000’s led to reduce the number of meals consumed at home in favor of stre et food in urban settings 145 . Research suggests that in low-income households, meat is the food category most restricted in the children’s diet 277 . Moreover, reduced supervision by mothers and caretakers at school age, and increased autonomy for feeding may also partly explain the degradation of dietary intake and nutritional status all along late childhood and adolescence. Indeed, without any public school feeding program in urban areas, children reaching the school age are mostly self-catered, which may cause poor food habits. As in many developing countries, school lunches in Senegal are limited to areas of high food insecurity, in rural areas, or where the World Food Program provides school meals.

Possible interventions

Because of its implication in impaired physical and cognitive development during school age and adolescence 17 , undernutrition associated with insufficient energy, protein and micronutrients intake in children from Senegalese urban schools should be addressed. School may be an efficient framework for nutritional and food interventions such as school feeding, nutrition education, or supplementation. A non-negligible part of the foods consumed by children and adolescent in urban settings is street food, therefore nutritional education programs should encourage the purchase of fruits instead of low-quality snacks 145 . Accessible and cheap dark green leafy vegetables and legumes are sufficient to ensure adequate folic acid status in poor populations 278 . Although bioavailability of vitamin A is much higher in animal products, these are also less affordable so carotenoids are usually the main source of vitamin A for economically deprived populations 85 . Consequently, traditional Senegalese recipes

69 containing unrefined palm oil and green leafy vegetables should be promoted among mothers in order to improve vitamin A and folic acid status. In African settings, school feeding program was shown to improve intake of energy 279 and micronutrients 151,237 , such as calcium, vitamin C and vitamin A through the distribution of fruits or dairy products. Beneficial effects of fortified food distribution through school feeding programs on micronutrient status were reported 175,280 . Calcium supplementation showed positive effects on calcium bone acquisition of children accustomed to low-calcium diets 281,282 and on long-term bone mass growth 283 .

Strengths and limitations

24-hours recall can be as efficient as weighted records 284 and it is more accurate than food frequency questionnaire to evaluate nutrient intake among school age children as reported in a review 285 , especially when it is repeated at least once 286 . It is statistically more efficient to increase the number of individuals than the number of days of recall, but at least more than one day is needed to estimate the intra-individual variability 287 . It is admitted that children from 7- 8 years are able to recall food consumed during the previous 24-hours recall but that for children under 7 years, parental help is needed 288 . However, dietary recalls collected from mothers can be inaccurate as out-of-home food intake is often omitted from mother’s recall 289 . Plus, in the present study, it was not possible to interview children along with their mothers because of their out-door morning activities but they were asked to check the recall of their child when the interviewers came to the household to collect recipes. Among low-income households, fluctuations in food supplies especially low before paydays makes diets highly variable 277 , which was taken into account in the present study conducted over 30 days. In Senegal, home-cooked meals are usually eaten in a common plate which complicates the estimation of individual’s consumption of staple food. The use of salted replicas to estimate rice portions in the present study was likely to be more accurate than the number of handfuls, which was reported as not satisfactory in Senegalese young children 290 . For other food, the use of handfuls, households measures and photographs may have lacked of accuracy 288 . But authors assume that the purchase price was a good indicator for children and mothers with limited budgets, likely to remember well the prices of street food and ingredients. The food composition table was derived from 5 tables, in which composition of prepared meals or snacks in Senegal was not available, so we gathered quantitative recipes for all households and street vendors. Home-cooked meals composition was specific to each child for each day, and the composition of snacks sold in the school or in the surroundings was specific to each school. Therefore, one of the strengths of the study is that nutrient composition of food was not approximated and that inter-households, intra-areas and day-to-day variability were taken into account. Albeit its infrequent use, the EAR 291 is still considered by some peers as the most relevant Dietary Reference Intake to assess risk of insufficient nutrient intake at population level, which takes into account the variability for both the requirement and the usual intake 233 . This method limits over- and underestimation of the prevalence 292 as it is observed in the present study and in other research 291 . However, the EAR cut-point may not be adapted to energy intake and to iron intake, due to the high correlation between intakes and energy requirements, and to the high skewness of iron requirement distribution in menstruating women 293 . But only 38 girls in the present sample had already reached menarche, representing 7% of the whole sample. Overall, the main recommendations for using the EAR cut-point method to evaluate adequacy of nutrient intake in a group was followed: minimum 100 individuals, to adjust for day-to-day variability and to exclude under-reporters 294 . One may object that EAR, AI and UL used in the

70 current study were based on the diet of healthy individuals from North America, which may not be adapted to African populations. However, the poor micronutrient status indicated by the biochemical data supports high prevalence of insufficient nutrient intake found in the present paper.

Conclusion

Diets were energy-deficient and unbalanced between poor protein and fiber and excessive fat and SFA. For all micronutrients, at least half of the children had insufficient intake, suggesting a diet poor in dairy products, meat, fruit and vegetables, with a special concern for zinc, vitamin A, folic acid and calcium. Nutrient intake inadequacy increases with age. These findings are in accordance with anthropometric and biochemical data indicating high rates of thinness and micronutrient deficiencies. Similarly to what was observed in other African rural and urban populations of school-aged children and adolescents, nutrition of school-aged children and adolescents from primary state schools of Dakar is worrisome, because of high nutrient needs related to growth spurt, puberty, school achievement or early pregnancies in adolescent girls. Reduction of familial care, increase of street food consumption, conjectural food prices crisis may explain a poorer diet at school age compared to infancy, which seems to be aggravated at adolescence. The findings of the present study highlight the need of nutritional interventions in Senegalese urban schools such as school feeding program and nutrition education.

Chapter 3. Evaluation of determinants and consequences of malnutrition among school children: data from Cambodia case

3.1. Stunting, poor iron status and parasite infection are significant risk factors for lower cognitive performance in Cambodian school children (Published paper)

Published in Plos One (2014)

Marlene Perignon a, Marion Fiorentino a, Khov Kuong e, Kurt Burja f, Megan Parker g, Sek Sisokhom h, Chhoun Chamnan e, Jacques Berger a, Frank T Wieringa a

Introduction

Worldwide, undernutrition and micronutrient deficiencies substantially impair human health and socio-economic development. Both developed and developing countries are concerned by the burden of micronutrient deficiencies disorders, although the highest prevalence exist in Sub-Saharan Africa and South Asia 3. In Cambodia, despite a considerable reduction in national poverty since the mid-1990s, undernutrition remains a major problem, and approximately one fifth of the population was still living on less than $1.25 per day in 2009. According to the most recent Cambodia Demographic and Health Survey (CDHS 2010) , 40% of children are stunted, 11% are wasted, and 55% of children of 6-59 months are anemic 295 .

Nutrition is one of many key factors affecting mental development of children. Deficiencies of critical micronutrients such as iron, folate or iodine can lead to impaired cognitive functions due to their decisive role in brain development. For example, iron deficiency (ID) has been estimated to impair the optimal mental development of 40% to 60% of the developing world’s infants 296 . Indirectly, deficiencies of other micronutrients like vitamin A or zinc can increase the risk of morbidity thereby increasing school absenteeism and reducing learning abilities and school performance.

Iron deficiency anemia (IDA), the most severe degree of ID, has been associated with poorer cognitive performance but it is still unclear if malnutrition has detrimental effects on cognitive performance independent of anemia in school-aged children 51 . Some recent systematic reviews based on randomized controlled trials provided evidence for a positive effect of iron supplementation on different measures of cognition in anemic and non-anemic children >5y 55 , on attention and concentration in adolescents and women 297 , and beneficial effect of micronutrient interventions (food-based or supplementation) on short term memory 298 . In addition to IDA, stunting is a known risk factor for impaired child development 299 . Despite a general decrease of the global prevalence, stunting still affects one third of the children under 5 y in the developing world 300 . Several studies have documented the relationship between cognitive ability and stunting in young children 301,302 but few have explored it within school- aged children 17,303,304 . School-aged children are often omitted from public health research. Early child development programs typically focus on the nutritional status and milestones of children under 5 y. Thus, our knowledge regarding the prevalence and impact of micronutrient deficiencies on children >5 y remains scarce. Yet, some areas of the brain and higher cognitive functions continue to develop throughout childhood and adolescence 36 . The myelination of frontal lobes which are thought to be responsible for executive, “higher -order” cognitive activities, starts around 6mo and continues until adulthood. Changes in the volumes of cortical gray and white matters, significantly correlated to children’s performance on a verbal learning task, occur during childhood and adolescence 305 . As during infancy, this ongoing brain development throughout

74 childhood is likely to be affected by detrimental effect of poor nutritional status, hence warranting further research on the impact of nutritional status on cognitive performance of school-aged children. In Cambodia, there is currently no data available concerning the nutritional status of children >5y, nor how malnutrition potentially affects their mental development. The present study aimed to evaluate the anthropometric and micronutrient status (iron, vitamin A, zinc and iodine) of Cambodian children aged 6-16 y and to determine if these outcomes are associated with cognitive performance.

Subjects and methods

Study population and design

Data were collected as part of a randomized placebo-controlled trial investigating the impact of multi-micronutrient fortified rices on health and development of Cambodian schoolchildren (FORISCA UltraRice+NutriRice study, FOrtifiedRIce for School meals in CAmbodia). Baseline data collection was conducted in November 2012 in 20 primary schools from 5 districts of Kampong Speu province in Cambodia. Kampong Speu is one of the 23 provinces of Cambodia, situated 60 km west of the capital Phnom Penh. Agriculture is predominant, with rice farming as the main occupation and income source. The schools were randomly selected from primary schools participating in school meal or take-home ration programs of the UN World Food Program (WFP). Children attending the selected schools were eligible to be part of the study if they were between 6-16 y of age, had written informed consent from parents/caregivers and did not have any mental or severe physical handicap. In each school, 132 children were randomly selected after stratification by sex and grade, hence 2640 children. 197 children were not recruited because they were absent on the day of data collection or refused to participate. Hence, a total of 2443 schoolchildren aged 6-16 y participated in the study. The study was approved by the National Ethic Committee for Health Research (NECHR) of the Ministry of Health, Phnom Penh, Cambodia, the Ministry of Education, Youth and Sports, Phnom Penh, Cambodia, and the Research Ethics Committee of PATH, Seattle, USA.

Anthropometric measurements

Height and weight were measured using standardized procedures 18,159 . Children wore minimum clothing and no shoes. Height was measured in duplicate to the nearest 0.1 cm with a wooden stadiometer. Weight was measured once, to the nearest 0.1 kilogram using an electronic scale (Seca, 881 U, Germany). Z-scores for height-for-age (HAZ) were calculated with AnthroPlus software version 1.0.4 using the WHO 2007 standards. Moderate and severe stunting were respectively defined as -3

Cognitive performance tests

The cognitive performance evaluation included thre e tests: the Raven’s Colored Progressive Matrices test (RCPM), and two standardized tests from the Wechsler Intelligence Scale for Children (WISC III): block design and picture completion. RCPM, the coloured form of Raven’s Progressive Matrices test for use with children, is a widely used nonverbal test of intelligence which was designed as a measure of overall intellectual ability 37,306 . The WISC III, designed for

75 children aged 6-16 y, is one of the most widely used tests of the intelligence of children 307 . Block design is a measure of problem solving to assess executive functions 307 . Picture completion evaluates alertness to detail and visual discrimination. The cognitive tests were conducted by 25 students from Psychology department of Royal University of Phnom Penh. The interviewers were trained in a 1 week workshop to ensure standardization in the assessment and scoring procedures. Each child was tested individually using standardized test protocols translated into Khmer language. Since norms are not available for Cambodia, interpretations of the scores of cognitive test were used as raw scores. For all cognitive tests, higher scores indicatebetter performance.

Blood and urine samples collection

Blood samples (5ml) were collected by venipuncture and aliquoted in a trace-element free vacutainers with no anticoagulant (Vacuette, Greiner Bio One, Austria). Urine samples were collected from the children in a sterile plastic container. Blood and urine samples were then stored in cool-boxes containing ice-packs and transported to Phnom Penh within 5h of collection. The blood samples were centrifuged at 2700 rpm (1300g) for 10 min at room temperature. Serum and urine were then aliquoted in capped Eppendorf tubes and stored at - 30°C until transfer for analysis.

Hemoglobin concentration

Hemoglobin concentrations were determined immediately after blood taking using the HemoCue (301+ system, HemoCueAngholm, Sweden). Anemia was defined as hemoglobin concentration <115 g/L for children between the ages 6 and 11 y, <120 g/L for children between the ages 12 and 14 y and girls aged 15 y and older and <130 g/L for boys aged 15 y and older according to WHO guidelines 308 .

Laboratory analysis

Ferritin (FER), soluble transferrin Receptor (TFR), retinol-binding protein (RBP), C-reactive protein (CRP), α1 -acid-glycoprotein (AGP) and zinc serum concentrations Serum samples were sent on dry ice to the VitMin laboratory (Willstaett, Germany) for determination of retinol-binding protein (RBP), C-reactive protein (CRP), ferritin (FER), soluble transferrin receptor (TFR ) and α1 -acid-glycoprotein (AGP), and to National Institute of Nutrition (Hanoi, Vietnam) for zinc analysis. RBP, FER, TFR, CRP, AGP were measured by a sandwich enzyme-linked immunosorbent assay (ELISA) technique 195 . Zinc concentration was measured using a flame atomic absorption spectrophotometer (GBC, Avanta+) using trace element-free procedures. Inflammation was defined as high CRP (>5mg/L) and/or high AGP concentrations (>1g/L). Inflammation status was then categorized in four groups based on CRP and AGP levels: no inflammation (normal CRP and AGP), incubation (high CRP and normal AGP), early convalescence (high CRP and AGP), and late convalescence (normal CRP and high AGP) 197 . FER is affected by presence of infection or inflammation, therefore FER concentrations were adjusted using correction factors published by Thurnham et al., namely 0.77, 0.53 and 0.75 for children respectively in incubation, early convalescence, and late convalescence phases. Low FER (corrected value<15 µg/L) was used as indicator of depleted iron stores 308 , and high TFR (>8.3mg/L) as indicator of tissue iron deficiency 210,309 . Iron deficiency was defined using both

FER and TFR indicators, i.e. by depleted iron stores or tissue iron deficiency. Total body iron (BI) was calculated from FER corrected for inflammation and TFR as described by Cook et al. 199 . Serum retinol is bound to RBP in a 1-to-1 complex, hence RBP concentrations were used to evaluate vitamin A status 200 . RBP concentrations were adjusted for the presence of inflammation using correction factors of 1.15, 1.32 and 1.12 for incubation, early convalescence and late convalescence phases respectively 201 . Corrected RBP cut-offs were used to define marginal vitamin A status (0.70 μmol/L ≤ corrected RBP < 1.05 μmol/L), vitamin A deficiency (< 0.70 μmol/L) and severe vitamin A deficiency (<0.35 μmol/L) respectively 200,310 . Zinc deficiency was defined using the following cut-offs: serum zinc concentration < 0.66 mg/L for girls ≥ 10 y, <0.70 mg/L for boys ≥ 10 y, and <0.65 mg/L for boys and girls <10 y 311 .

Urinary iodine concentration Urine samples were sent on dry ice to the Provincial Preventive Medicine Center (Thai Nguyen, Vietnam) for determination of urinary iodine concentration (UIC) using spectrophotometric methods 312 . Iodine deficiency (IDD) was defined by a median UIC below 100 µg/L and/or a proportion of participants below 50 µg/L higher than 20%, and iodine nutrition above requirements by a median UIC above 200 µg/L 313 .

Parasite infestation Plastic containers and instructions for fecal sample collection were distributed to the children on the day of data collection and requested to be returned with fecal sample to the school the following day. Samples were then stored in a cool box, transported to the National Malaria Center (CNM, Phnom Penh, Cambodia) and stored at 4°C until analysis. Quantitative parasite egg counts were performed by CNM using the Kato-Katz method 314 . The egg output was expressed as eggs per gram feces (epg).

Socio-economic survey

Socio-economic information was collected on a sub-sample (n= 616 children, FORISCA- NutriRice study) by trained interviewers during household visits. Questionnaires were answered by parents or caretaker and included information about household characteristics, caretaker’s level of education and amount and source of household income.

Data management and Statistical analysis

Data entry, including quality checks and validation by double entry of questionnaires, was performed with EpiData version 3.1 (EpiData Association, Odense, Denmark). Data management and analyses were performed using SPSS version 20.0 software (SPSS, Inc., Chicago, IL). Normality of distributions was evaluated using Kolmogorov-Smirnoff test. Not normally distributed data were considered to be reasonably close to normality to allow parametric tests when skewness and kurtosis values ranged between -1.0 and +1.0. Continuous variables that were not normally distributed were log-transformed. Block design score was categorized as score below and above median value. All analyses took into account characteristics of the cluster sampling design (school clusters). The main effects of explanatory variables on cognitive tests scores were first assessed in univariate analysis using ANCOVA for RCPM and Picture completion tests, and logistic regression for Block design test, with all analysis including age as covariate. Multivariate analyses were then performed to evaluate the associations between

77 cognitive tests scores and variables assessing nutritional status (stunting, anemia and micronutrient status) while taking into account the effect modification or confounding of other variables (age, gender, parasite infection, socio-economic status). Any variable having a p-value <0.25 in the univariate test was considered for the multivariate analysis. Interactions were tested and if significant at the 0.05 level, analyses were run separately at each level of the variable modifying the effect. Multiple comparisons were conducted by using the Bonferroni post-hoc test.

Results

Characteristics of the studied sample

A total of 2443 children from grades 1-6 participated in the study. Characteristics of the participants are presented in Table 13. The mean ± SD age of children was 9.6 ± 2.3 y and half (49.9%) were girls. The prevalence of anemia was 15.7% including 0.1% severe anemia. There was no significant difference in anemia prevalence between boys and girls, or age groups; however, prevalence varied between 4.8% and 30.0% across the schools. The overall prevalence of stunting (HAZ<-2SD) was 40.0%, including 10.9% severe stunting (HAZ<-3SD). Children ≥10 y were significantly more affected by stunting (54.0% and 29.6% respectively) or severe stunting (18.5% and 5.3% respectively) than children<10 y, but no difference was found between gender. Inflammation (CRP>5 mg/L and/or AGP>1 g/L) was found in more than one third of the children (39.5%), with boys being more affected than girls. Parasite infection was found in 18% of the children, boys being more affected than girls (p<0.05). Only 1.5% of the children had depleted iron stores (FER<15 µg/L) whereas 50.1% had tissue iron deficiency (TFR>8.3 mg/L). There was no significant difference between age groups or genders for both indicators. Prevalence of iron deficiency, as defined by low FER and/or high TFR, was 51.2% without any difference between age groups or gender. Only 2% of the children had negative body iron stores but marginal body iron stores (total body iron<4 mg/kg body weight) were prevalent (13.9%). Prevalence of iron-deficiency anemia was 10.0% with no difference between boys and girls. Prevalence of vitamin A deficiency (VAD) was 0.7%, with no severe VAD. However, 7.9% of the children had marginal VA status (0.7≤corrected RBP<1 .05 µmol/L). Children <10 y were significantly more affected by marginal VA status (9.4% and 6% respectively) than children >10 y (p<0.05). Prevalence of marginal VA status was also higher in boys than in girls (9.2% and 6.6% respectively, p<0.05). Most children (93%) exhibited low serum zinc concentrations, indicative of zinc deficiency. Approximately one fifth (17%) of the children were iodine deficient while half (50.2%) had iodine intake above requirements.

TABLE 13 CHARACTERISTICS OF SCHOOL -CHILDREN PARTICIPATING IN THE STUDY

BOYS GIRLS ALL p-value n 1223 1220 2443 Age 9.75 ± 2.34 9.54 ± 2.17 9.65 ± 2.26 p<0.05 % inflammation 41.6 (n=491) 37.5 (n=446) 39.5 (n=937) p<0.05 % parasite infection 20.0 (n=177) 16.1 (n=148) 18.0 (n=325) p<0.05 ANTHROPOMETRY HAZ -1.80 ± 1.00 (n=1219) -1.71 ± 1.06 (n=1216) -1.75 ± 1.03 (n=2435) p<0.05 % HAZ<-2SD 41.6(n=507) 38.5 (n=468) 40.0 (n=965) NS % HAZ<-3SD 11.1(n=135) 10.8 (n=131) 10.9 (n=266) NS IRON STATUS Hb (g/L) 123.8 ± 9.9 (n=1203) 124.5 ± 9.5 (n=1206) 124.2 ± 9.7 (n=2409) NS % anemia 16.8 (n=202) 14.7 (n=177) 15.7 (n=379) NS % severe anemia 0.1 (n=1) 0.2 (n=2) 0.1 (n=3) FER 1,2 (mg/L) 66.28 ± 1.68 (n=1181) 68.63 ± 1.68 (n=1190) 67.45 ± 1.68 (n=2371) NS % FER 1 <15 mg/L 1.9 (n=22) 1.2 (n=14) 1.5 (n=36) NS TfR 2 (mg/L) 8.56 ± 1.32 (n=1181) 8.41 ± 1.30 (n=1190) 8.48 ± 1.31 (n=2371) NS % TfR >8.3 mg/L 51.8 (n=612) 50.3 (n=598) 51.0 (n=1210) NS % ID 3 total 52.2 (n=616) 50.3 (n=599) 51.2 (n=1215) NS % ID 3 with anemia 10.8 (n=127) 9.3 (n=111) 10.0 (n=238) NS Body iron (mg/kg) 5.90 ± 2.27 (n=1181) 6.09 ± 2.21 (n=1190) 5.99 ± 2.24 (n=2371) p<0.05 % body iron <0 2.5 (n=29) 1.6 (n=19) 2.0 (n=48) NS % body iron <2 5.0 (n=59) 5.0 (n=59) 5.0 (n=118) NS % body iron <4 14.5 (n=171) 13.3 (n=158) 13.9 (n=329) NS VITAMIN A STATUS RBP 1 (µmol/L) 1.54 ± 0.41 (n=1181) 1.62 ± 0.45 (n=1190) 1.58 ± 0.43 (n=2371) p<0.01 % marginal VA status 4 9.2 (n=109) 6.6 (n=79) 7.9 (n=188) p<0.05 % VAD 5 0.8 (n=10) 0.6 (n=7) 0.7 (n=17) NS IODINE STATUS % iodine deficiency 6 14.8 (n=175) 19.8 (n=233) 17.3 (n=408) p<0.05 % above requirements 7 52.5 (n=620) 47.9 (n=564) 50.2 (n=1184) p<0.05 ZINC STATUS % zinc deficiency 8 93.7 (n=931) 91.9 (n=907) 92.8 (n=1838) NS SOCIO-ECONOMIC STATUS (on a sub-group n=616) Income 9 ($/year) 1995 (495;4291) 1825 (500;4050) 1935 (500;4150) Caretaker's level of education: % no or informal schooling 11.9 (n=35) 14.8 (n=42) 13.3 (n=77) NS % primary school 65.3 (n=192) 59.0 (n=167) 62.2 (n=359) NS % secondary school 22.8 (n=67) 26.1 (n=74) 24.4 (n=141) NS

Results are mean ± SD unless stated, 1corrected for inflammation, 2geometric mean ± SD, 3based on FER 1 < 15 mg/L and/or TFR> 8.3 mg/L, 4 0.7 ≤ RBP 1< 1.05 µmol/L, 5RBP 1 < 0.7 µmol/L, 6UIC<100 mg/L, 7UIC ≥ 200 mg/L, 8zinc < 0.66 mg/L for girls ≥ 10 y, <0.70 mg/L for boys ≥ 10 y, and <0.65 mg/L for boys and girls <10 y, 9median (10th;90th ), HAZ: height-for-age z-scores, Hb: hemoglobin, FER: ferritin, TFR: transferrin receptors, ID: iron-deficiency, NS: not significant, RBP: retinol binding protein, VA: vitamin A, VAD: vitamin A deficiency

Factors associated with cognitive performance

Cognitive scores were associated with stunting and micronutrient status (Table 14 (RCPM test), Table 15 (Picture completion test) and Table 16 (Block design test)). Children with severe stunting scored significantly lower than non stunted children in all tests (p<0.001 for all), with reduction of score reaching up to -2 points in RCPM test (12% of mean score), or -1.7 points in picture completion test (22% of mean score) after adjustment on all variables. Children with moderate stunting scored significantly lower (Picture completion and Block design tests, both p<0.001) or tend to score lower (RCPM test, p=0.053) when compared with non-stunted children. For the block design test, severely stunted children were 2.5 times more likely to score below the median value than children with normal height status (OR=2.53, p<0.001). Univariate analysis showed that posit ive but marginal body iron stores (0

TABLE 14 FACTORS ASSOCIATED WITH COGNITIVE PERFORMANCE IN RCPM TEST AMONG PARTICIPATING SCHOOL CHILDREN

Estimated mean Estimated mean Estimated difference (95% CI) Variable n difference 1 p-value p-value mean 1 ± SE adjusted for all (95% CI) variables 9 Gender male 1201 16.88 ± 0.14 female 1224 16.44 ± 0.14 -0.44 (-0.77; -0.12) 0.008 -0.72 (-1.12; -0.33) p<0.001 Stunting normal status 1500 17.03 ± 0.13 -3 < HAZ < -2 677 16.41 ± 0.26 -0.62 (-1.32; 0.08) 0.10 -0.56 (-1.12; 0.006) 0.053 HAZ < -3 244 15.12 ± 0.29 -1.91 (-2.70; -1.12) p<0.001 -1.97 (-2.81; -1.12) p<0.001 Parasite infection not infected 1471 17.49 ± 0.11 infected 324 17.00 ± 0.28 -0.49 (-1.09; 0.11) 0.11 M 0.29 (-0.49; 1.08) 0.47 F -1.43 (-2.18; -0.68) 0.016 Iron deficiency 2 no ID 1148 17.21 ± 0.13 ID without anemia 968 16.93 ± 0.14 -0.28 (-0.73; 0.18) 0.43 M -0.31 (-1.10; 0.48) 1.00 F -0.31 (-1.01; 0.38) 0.84 ID with anemia 236 16.41 ± 0.35 -0.81 (-1.70; 0.086) 0.092 M -1.46 (-2.73; -0.18) 0.019 F -0.36 (-1.56; 0.84) 1.00 FER 3 (mg/L) ≥50 1811 17.19 ± 0.10 15-50 507 16.69 ± 0.20 -0.50 (-1.03; 0.02) 0.068 <15 34 16.50 ± 1.07 -0.69 (-3.26; 1.87) 1.00 TfR (mg/L) ≤8.3 1152 17.21 ± 0.13 >8.3 1200 16.86 ± 0.12 -0.348 (-0.698; 0.01) 0.052 Body iron (mg/kg) >4 2028 17.16 ± 0.09 0-4 278 16.42 ± 0.28 -0.75 (-1.44; -0.06) 0.029 <0 46 15.85 ± 0.82 -1.32 (-3.29; 0.65) 0.33 VA status normal VA status 2170 16.69 ± 0.12 marginal VA status 4 173 16.48 ± 0.42 -0.21 (-1.06; 0.65) 0.63 VAD 5 15 16.15 ± 1.26 -0.54 (-3.02; 1.95) 0.67 Iodine status adequate 763 17.21 ± 0.18 iodine deficiency 6 406 16.72 ± 0.33 -0.49 (-1.39; 0.41) 0.57 above requirements 7 1171 17.33 ± 0.14 1.12 (-0.42; 0.65) 1.00 Zinc status normal status 140 17.62 ± 0.51 zinc deficiency 8 1827 17.08 ± 0.11 -0.54 (-1.57; 0.49) 0.30 1adjusted for age, 2defined by FER 3< 15 mg/L and/or TFR > 8.3 mg/L, 3corrected for inflammation, 40.7 ≤ RBP 3< 1.05 µmol/L, 5RBP 3< 0.7 µmol/L, 6UIC<100mg/L, 7UI C ≥ 200mg/L, 8zinc < 0.66 mg/L for girls ≥ 10 y, <0.70 mg/L for boys ≥ 10 y, and <0.65 mg/L for boys and girls <10 y, 9significant interactions between gender and parasite, and gender and iron deficiency, ID: iron deficiency, FER: ferritin, TFR: transferrin receptors, VA: vitamin A, VAD: VA deficiency, M: male, F: female, RCPM: Raven’s colored progressive matrices

TABLE 15 FACTORS ASSOCIATED WITH COGNITIVE PERFORMANCE IN PICTURE COMPLETION TEST AMONG PARTICIPATING SCHOOL CHILDREN

Estimated mean Estimated Estimated mean difference (95% CI) Variable n mean 1 difference 1 p-value p-value adjusted for all ± SE (95% CI) variables 9 Gender male 1201 7.38 ± 0.13 female 1224 7.29 ± 0.14 -0.09 (-0.41; 0.23) 0.60 Stunting normal status 1500 7.80 ± 0.13 -3 < HAZ < -2 677 6.69 ± 0.25 -1.11 (-1.79; -0.43) p<0.001 -1.07 (-1.62; -0.52) p<0.001 HAZ < -3 244 5.99 ± 0.29 -1.81 (-2.58; -1.04) p<0.001 -1.68 (-2.50; -0.86) p<0.001 Parasite infection not infected 1485 7.69 ± 0.18 infected 304 7.09 ± 0.31 -0.60 (-1.30; 0.10) 0.095 M 0.02 (-0.73; 0.77) 0.96 F -0.85 (-1.61; -0.09) 0.028 Iron deficiency 2 no ID 1148 7.43 ± 0.17 ID without anemia 968 7.24 ± 0.18 -0.19 (-0.78; 0.40) 1.00 -0.49 (-1.01; 0.02) 0.064 ID with anemia 236 7.00 ± 0.36 -0.43 (-1.39; 0.52) 0.82 -0.81 (-1.66; 0.04) 0.067 FER 3 (mg/L) ≥50 1811 7.89 ± 0.10 15-50 507 7.53 ± 0.19 -0.36 (-0.88; 0.15) 0.27 <15 34 8.11 ± 1.05 0.22 (-2.31; 2.74) 1.00 TfR (mg/L) ≤8.3 1152 7.98 ± 0.12 >8.3 1200 7.64 ± 0.12 -0.35 (-0.69; -0.01) 0.048 Body iron (mg/kg) >4 2028 7.90 ± 0.09 0-4 278 7.13 ± 0.27 -0.77 (-1.45; -0.10) 0.019 <0 46 7.33 ± 0.80 -0.57 (-2.51; 1.36) 1.00 VA status normal VA status 2170 7.36 ± 0.11 marginal VA status 4 173 7.56 ± 0.41 0.19 (-0.82; 1.21) 1.00 0.08 (-0.84; 0.99) 1.00 VAD 5 15 4.79 ± 1.22 -2.57 (-5.52; 0.37) 0.11 -0.86 (-3.43; 1.71) 1.00 Iodine status adequate 763 7.73 ± 0.17 iodine deficiency 6 406 7.72 ± 0.33 -0.01 (-0.90; 0.88) 1.00 above requirements 7 1171 8.03 ± 0.14 0.30 (-0.23; 0.83) 0.51 Zinc status normal status 140 8.21 ± 0.50 zinc deficiency 8 1827 7.83 ± 0.10 -0.38 (-1.38; 0.62) 0.45

1adjusted for age, 2defined by FER 3< 15 mg/L and/or TFR > 8.3 mg/L, 3corrected for inflammation, 40.7 ≤ RBP 3< 1.05 µmol/L, 5RBP 3< 0.7 µmol/L, 6UIC<100mg/L, 7UIC ≥ 200mg/L, 8zinc < 0.66 mg/L for girls ≥ 10 y, <0.70 mg/L for boys ≥ 10 y, and <0.65 mg/L for boys and girls <10 y, 9significant interactions between gender and parasite, ID: iron deficiency, FER: ferritin, TFR: transferrin receptors, VA: vitamin A, VAD: vitamin A deficiency, M: male, F: female

TABLE 16 UNIVARIATE AND MULTIVARIATE ANALYSIS OF FACTORS ASSOCIATED WITH POOR PERFORMANCE IN BLOCK DESIGN TEST AMONG PARTICIPATING SCHOOLCHILDREN

odds-ratio odds-ratio 1 Variable n p-value adjusted for all p-value (95% CI) variables (95% CI) Gender male 1208 1 1 female 1213 1.28 (1.06; 1.55) 0.009 1.27 (0.99; 1.63) 0.058 Stunting normal status 1446 1 1 -3 < HAZ < -2 706 1.88 (1.50; 2.37) p<0.001 1.73 (1.29; 2.33) p<0.001 HAZ < -3 263 3.67 (2.56; 5.26) p<0.001 2.35 (1.47; 3.75) p<0.001 Parasite infection not infected 1471 1 1 infected 324 1.64 (1.21; 2.24) 0.002 1.73 (1.24; 2.42) 0.001 Iron deficiency 2 no ID 1148 1 1 ID without anemia 968 0.95 (0.77; 1.17) 0.62 1.09 (0.83; 1.43) 0.52 ID with anemia 236 1.32 (0.95; 1.85) 0.10 1.17 (0.74; 1.85) 0.26 FER 3 (mg/L) ≥50 1811 1 15-50 507 1.11 (0.88; 1.40) 0.39 <15 34 1.31 (0.60; 2.90) 0.50 TfR (mg/L) ≤8.3 1152 1 >8.3 1200 1.01 (0.83; 1.23) 0.93 Body iron (mg/kg) >4 2028 1 0-4 278 1.23 (0.92; 1.66) 0.17 <0 46 1.24 (0.62; 2.49) 0.54 VA status normal VA status 2149 1 marginal VA status 4 186 0.91 (0.64; 1.30) 0.60 VAD 5 17 1.67 (0.48; 5.89) 0.42 Iodine status adequate 763 1 iodine deficiency 6 406 0.91 (0.68; 1.21) 0.52 above requirements 7 1171 0.94 (0.74; 1.19) 0.60 Zinc status normal status 140 1 1 zinc deficiency 8 1827 1.43 (0.95; 2.15) 0.091 1.32 (0.82; 2.13) 0.26

1adjusted for age, 2defined by FER 3< 15 mg/L and/or TFR > 8.3 mg/L, 3corrected for inflammation, 40.7 ≤ RBP 3< 1.05 µmol/L, 5RBP 3< 0.7 µmol/L, 6UIC<100mg/L, 7UIC ≥ 200mg/L, 8zinc < 0.66 mg/L for girls ≥ 10 y, <0.70 mg/L for boys ≥ 10 y, and <0.65 mg/L for boys and girls <10 y, ID: iron deficiency, FER: ferritin, TFR: transferrin receptors, VA: vitamin A, VAD: vitamin A deficiency, M: male, F: female

TABLE 17 FACTORS ASSOCIATED WITH COGNITIVE PERFORMANCE IN RCPM TEST BEFORE AND AFTER ADJUSTMENT ON SOCIO -ECONOMIC STATUS IN A SUB -SAMPLE OF SCHOOL CHILDREN (N=616)

mean difference 2, 3 mean difference 1 Variable n p-value (95% CI) after p-value (95% CI) adjustment for SES Gender male 311 female 305 -0.29 (-1.06; 0.48) 0.47 -0.47 (-1.27; 0.33) 0.25 Stunting normal status 367 -3 < HAZ < -2 176 -0.33 (-1.42; 0.75) 1.00 -0.55 (-1.67; 0.57) 0.71 HAZ < -3 67 -1.68 (-3.33; -0.04) 0.044 -1.60 (-3.29; 0.08) 0.067 Parasite infection not infected 387 infected 48 M -0.09 (-1.85; 1.66) 0.92 M 0.20 (-1.63; 2.03) 0.83 F -2.32 (-4.24; -0.40) 0.018 F -2.27 (-4.25; -0.29) 0.025 Iron deficiency 4 no ID 261 ID without anemia 262 M 0.13 (-1.49; 1.74) 1.00 M -0.01 (-1.63; 1.65) 1.00 F -0.25 (-1.75; 1.24) 1.00 F -0.26 (-1.77; 1.25) 1.00 ID with anemia 65 M -2.09 (-4.60; 0.42) 0.14 M -2.23 (-4.78; 0.32) 0.11 F -1.05 (-3.25; 1.15) 0.76 F -0.95 (-3.31; 1.41) 1.00 Socio-economic status Household income 5 616 0.03 (0.00; 0.05) 0.048 6 Caretaker’s education 616 0.18 (0.02; 0.34) 0.024

1variables in the model: age, gender, stunting, parasite infection, iron deficiency, 2variables in the model: age, gender, stunting, parasite infection, iron deficiency, caretaker's level of education, income, 3interaction between gender and parasite infection, and gender and iron-deficiency, 4defined by FER(corrected for inflammation) < 15 mg/L and/or TFR > 8.3 mg/L, 5x100 US$/year, 6years of schooling, ID: iron deficiency, SES: socio-economic status, M: male, F: female

Discussion

Within this large cross-sectional survey of Cambodian school children, both long-term nutritional deficits as reflected by stunting and current micronutrient status, especially iron status, significantly affected cognitive performance. According to the WHO classification system, the observed prevalence of anemia (15.7%) indicated a mild public health problem in Cambodia, with only 3 cases of severe anemia. The prevalence of depleted iron stores (1.5%) or negative body iron (2%) were both very low. In contrast, more than half of the children had high TFR concentrations. TFR reflects tissue iron needs but may have also been increased by factors like hemoglobinopathy or inflammation 315,316 . TFR levels were actually higher in children showing abnormal Hb type (HbE) or inflammation, but prevalence of high TFR was still important in the sub-sample of children with normal Hb profile (48.5%) or without inflammation (42.1%). Iron stores being adequate for most of the children (as indicated by the high ferritin concentrations), high TFR concentration suggests a functional tissue iron deficiency due to either an impaired release of iron from stores or impaired systems for transporting iron to target tissues.

The prevalence of iron-deficiency anemia (10%) represented about two-thirds of the overall anemia prevalence. Also, approximately one tenth (9%) of the children exhibited low VA status,

84 with 0.7% being deficient, and >90% had low serum zinc concentrations, indicating a very high prevalence of zinc deficiency in this population. Hence, this population of school children exhibited concurrent deficiencies of micronutrients, as reported previously for younger children in SE Asia 104,223 . Poor iron status was associated with lower cognitive ability, however the effect was modified by gender: children with IDA, especially boys, scored lower than iron-replete children. Interestingly, ID was associated with lower scores in the picture completion test even without anemia. Also, lower scores in two of the three tests were found for children with positive but marginal total body iron stores (0-4 mg/kg) as compared to children with replete stores. Thus, the findings from this study indicate that marginal iron status may have impaired cognitive performance before the onset of anemia. This evidence suggests that interventions are needed to improve iron status in school children even in the absence of IDA. Stunting is an indicator of early-life chronic malnutrition, with most of the growth deficit occurring in the first 2 y of life 317 . In the present study, stunting was a high risk factor for lower scores in all tests after controlling for age, gender and micronutrient status (iron, vitamin A, zinc, iodine), and also for socio-economic status in the sub-sample of children. Furthermore, this study showed that the prevalence of stunting, both moderate and severe, was higher in children ≥10 y than in children <10 y. One possible explanation could be that nutritional status of young children has improved considerably over the last decade in this area, but the national data reported a consistent 40% of stunting among children <5 y in 2005 and 2010 295,318 . Another explanation could be that growth faltering continues after 5 y of age. Thus, considering also the extremely high prevalence of zinc deficiency found in the studied 6-16 y children (93%), interventions like zinc fortification could possibly help to limit worsening of stunting and, as a consequence, of cognitive performance impairment. Children infected by parasites, especially girls, exhibited poorer cognitive performance than non infested children. Possible links between parasite infection and cognitive outcomes are reduced school attendance due to illness, loss of concentration, or through nutrition by affecting absorption of micronutrients such as iron and iodine 319,320 . In the present study, parasite infections remained a highly significant factor in the multivariate analysis which included anthropometry and micronutrient status indicators, suggesting an effect of parasite infection on cognition independent from nutrition. Therefore, the negative effect of parasite infection on cognition might be associated to higher absenteeism from school due to illness. Regardless of the underlying mechanisms, deworming programs could benefit school performance in Cambodian school children.

Although not available for all studied children, socio-economic status was included in analysis for a sub-sample of children (n=616). The results obtained from this sub-group were similar to the whole sample and remained unchanged after correcting for confounding effect of caretaker’s level of education and household’ s income. Thus, we can reasonably assume that associations between cognitive outcomes and variables reflecting nutritional status found for the 2443 children are not due to confounding effects of socio-economic status.

Thus, the present study showed that poor cognitive performance was multi-factorial, with nutritional factors as stunting and iron deficiency and non-nutritional factors such as parasite infection being implicated. These results are consistent with recent studies showing a relationship between cognitive abilities below average and low HAZ in school children in South East Asia 304 , and a beneficial effect or iron supplementation on attention, concentration and different measures of cognition 55,297 . Hence, as stunting and iron deficiency in school-aged

85 children reflect long-term and current nutritional status respectively, interventions at different periods of age could have a beneficial effect on cognitive abilities. Nutritional programs in early life to ensure good nutrition and prevent stunting before start of schooling are crucial. In addition, strategies at school-age like deworming, surveys for early recognition and prevention of iron deficiency before the onset of anemia, or interventions to prevent worsening of growth faltering should be considered to contribute to optimal cognitive development. Combining nutrition specific interventions, such as food fortification, with nutrition sensitive interventions, such as deworming, might have synergistic benefits.

3.2. Height, zinc and soil-transmitted helminthes infections in school children: a study in Cuba and Cambodia (Published paper)

Published in Nutrients (2015)

Brechje de Gier i, Liliane Mpabanzi j, Kim Vereecken j, Suzanne D. van der Werff i, Patrick C. D’Haese k, Marion Fiorentino a, Kuong Khov e, Marlene Perignon a, Chhoun Chamnan e, Jacques Berger a, Megan E. Parker g, Raquel Junco Díaz l, Fidel Angel Núñez m, Lázara Rojas Rivero m, Mariano Bonet Gorbea l, Colleen M. Doak i, Maiza Campos Ponce i, Frank T. Wieringa a, Katja Polman ij

Introduction

Height for age, expressed as z-scores of internationally accepted reference curves, is recommended by the World Health Organization (WHO) and the United Nations Children’s Fund (UNICEF) and Food and Agriculture Organization (FAO) as an indicator of chronic undernutrition 321 . Undernutrition can be caused by insufficient intake of macronutrients, micronutrients or both. Poor growth has been associated with insufficient intake and/or absorption of micronutrients 300 . An important micronutrient deficiency prevalent in many middle- and low-income countries is zinc deficiency, for which over 20% of the world’s population is estimated to be at risk 3. Zinc, a trace metal micronutrient, influences many physiological functions, among which growth 322,323 . Deficiency in zinc is recognized as a major cause of morbidity and mortality in developing countries 203,324 . Though generally accepted as a public health concern, documentation on zinc deficiency at the population level remains challenging, as there is no gold standard for the measurement of zinc levels 325,326 . To date, plasma/serum zinc concentration, dietary intake, and stunting prevalence are the best-known indicators of zinc deficiency 203 .

Infections with soil-transmitted helminths (STH) such as Ascaris lumbricoides , Trichuris trichiura and hookworm affect approximately a quarter of the world’s population, and the vast majority of these populations live in middle- and low-income countries in (sub)tropical regions 327 . STH infections have been associated with reduced height for age and stunting, and are strongly related to poverty 328,329 . Populations of these endemic regions often show a poor nutritional status 330 . Zinc deficiency and STH infections are thus likely to coexist in these areas. Moreover, several studies have suggested a role for zinc in susceptibility to STH infections 331,332 . Although the effects of zinc deficiency and STH infections on growth have both been widely studied, data on the association between zinc, STH infection and growth are scarce.

Poor nutritional status and STH infection are intricately linked, whereby STH infection can lead to malnutrition and malnutrition may increase susceptibility to STH infection 332 . Likewise, STH infections and poor nutritional status can affect growth, either independently or in combination. Economic development, population nutritional status, as well as STH species distributions vary greatly between STH endemic countries. For example, Cambodia remains a low-income country with a high prevalence of stunting despite considerable economic development and significant improvement in its population health conditions since the end of the civil war. Food insecurity is still a reality for many of its inhabitants, and, additionally, a high prevalence of STH infection has been reported, mostly by hookworm and A. lumbricoides 333 . In contrast, Cuba, which is also an STH endemic country, has a high development index and is categorized as an upper middle- income country. In Cuba, the epidemiological transition has firmly settled in and overweight

87 rather than underweight is currently a public health concern 334 . Estimates of zinc deficiency prevalence are not available for these countries. The present paper aimed at assessing the associations between height for age, zinc status and STH infections in school-aged children in these two different populations.

Methods

Study Population Cuba

A cross-sectional study within school-aged children was performed in 2009 in San Juan y Martínez, Pinar del Rio, a municipality in the West of Cuba. The municipality is situated in a rural mountainous area, which is endemic for STHs 335 . From 13 randomly selected schools, 1389 children were included in the study. Written informed consent was obtained from the parents or caretakers of each child. The study was approved by the ethical committees of the Institute of Tropical Medicine in Antwerp (Belgium), the Pedro Kourí Institute of Tropical Medicine and the National Institute for Hygiene, Epidemiology and Microbiology in Havana (Cuba).

Study Population Cambodia

Data from the baseline measurements of a randomized controlled trial on the effects of multiple-micronutrient-fortified rice on child nutrition and morbidity were used. The trial was conducted in rural Kampong Speu province, Cambodia, in November 2012. Children from 20 randomly selected schools were included ( N = 2471). All parents or caretakers were asked to sign an informed consent form. Ethical approval was obtained from the Cambodian Ministry of Health, Education and Planning and the Ethical Review board of PATH, USA.

Height for Age

Height measurements were performed to the nearest 0.1 cm by trained investigators using standard procedures. Age in months was calculated from the children’s birth date, retrieved via interviews and verified by school records and birth certificates (Cambodia). Height for age z- scores were calculated according to the WHO 2007 reference curves, using the WHO macro for SPSS 157 . Stunting was defined as height for age z-score below −2 SD. For analyses where age or height for age as continuous covariates were not linearly associated with the dependent variable, data were categorized. Cutoffs were chosen so that three categories of approximately equal group size were made. Because age and height for age ranges differed between both populations, the categories were defined differently per population. In the Cuban data, age was categorized as 4 to 7, ≥7 to 10 and ≥10 to 13 years old. Cuban height for age z-scores were categorized as <0, 0 –1 and >1 SD. In the Cambodian data, age was categorized as 5 to 10, ≥10 to 13 and ≥13 to 17 years old. Here, height for age z -scores were categorized as ≤−2, −2 to 0 and >0 SD.

Parasitology and Treatment

In both countries, one fresh stool sample was collected from each child. Stools were examined by the Kato-Katz technique (duplicate 25 mg smears) according to standard procedures to detect A. lumbricoides , T. trichiura , and/or hookworm 336 . Infection intensity was recorded as eggs per gram feces (epg) and classified according to WHO guidelines. STH positive children

88 received anthelminthic treatment: in Cuba, one single dose of 500 mg mebendazole, which has been evaluated and is the treatment of choice in Cuba 337 and in Cambodia, one single dose of 400 mg albendazole was given 338 .

Plasma Zinc and Inflammation

In Cambodia, zinc was measured in plasma. C-reactive protein (CRP) and alpha-1 acid glycoprotein (AGP) were measured alongside plasma zinc, in order to adjust for the effects of inflammation on plasma zinc concentrations. Plasma zinc and CRP and AGP were measured in 5 mL of venous blood, obtained from participants by venipuncture. Plasma zinc concentration was measured by flame atomic absorption spectrophotometry and verified against reference material at the National Institute for Nutrition in Hanoi, Vietnam. Deficiency was defined as plasma zinc below 9.9 µmol/Lfor children below the age of 10, below 10.1 µmol/L for girls age 10 and older and plasma zinc below 10.7 µmol/L for boys age 10 and older 203 . In 100 µL plasma aliquots, CRP and AGP were measured by sandwich enzyme-linked immunosorbent (ELISA) techniques (VitMin Laboratories, Germany) 195 . Inflammation categories were defined as elevated CRP only, elevated AGP only, both CRP and AGP elevated or no elevated CRP or AGP. Elevated CRP was defined as values above 5 mg/L, elevated AGP was defined as > 1 g/L 339 .

Hair Zinc

In Cuba, zinc was measured in hair. Two months before the measurements, parents or guardians of the participating children were asked not to cut the hair of their children. Approximately 200 –500 mg of hair was collected with the use of stainless steel scissors in the nape or (lower) occipital region of the head approximately 1.5 cm away the scalp. The distal ends of the hair were cut from the samples, leaving a specimen of approximately 2 cm in length. Samples were stored in plastic bagsat −20 °C until the determination of the zinc content. In the laboratory, the samples were analyzed for zinc content by spectrometry. In order to assure the quality of the zinc measurements taken, samples first underwent a washing procedure, to remove exogenous zinc without removing endogenous zinc. Ultra-pure reagents and pretested vials were used. Zinc analysis was done according to the protocol of D’Haese et al .340 . A cutoff value of 70 µg/g wet weight was used to define zinc deficiency 203,341 . Due to funding restraints, hair zinc was measured in a subset of 230 Cuban children.

Statistical Analysis

Analyses were done using SPSS software version 21 (IBM, NY, USA). Hair zinc followed a skewed distribution, therefore the data for this variable were natural log-transformed for regression analysis and expressed as median and interquartile range for descriptive analysis. The variable STH infection refers to the presence of any STH infection, ‘zinc’ refers to zinc concentration and ‘height for age’ refers to height for age z -score in all analyses. For statistical testing, linear regression analysis was performed with height for age z-scores, plasma zinc or the natural logarithm of hair zinc as continuous dependent variables. Covariates of each analysis are specified in the table footnotes. In the analyses of associations between zinc and STH infection with height for age, age was included as a continuous covariate and inflammation categories were included as categorical covariate for the plasma zinc data. In the analysis of associations between zinc and STH infection, covariates age and height for age z-scores were included as categorical variables, created from age and height for age categories. Sex was added

89 as binary covariate in all analyses. Statistical significance was defined as a p value below 0.05, for variables as well as interaction terms.

Results

Characteristics of the Study Populations

The mean height for age z-score (0.06) of the Cuban children was significantly higher than the median of the reference population (z-score = 0) (p 0.03). Only 21 (1.6%) of the Cuban children presented with stunting (Table 18). In the Cambodian children, mean height for age z-scores were significantly lower than 0 ( p< 0.001) and stunting was common (42.9%). Zinc deficiency was highly prevalent in Cambodia (92.8%), whereas zinc deficiency was found in only 12.2% of the Cuban children. Prevalence of STH infections was 8.4% and 16.8% for Cuba and Cambodia, respectively. In the Cuban study, the most common STH infections were A. lumbricoides (61.4%) and T. trichiura 36.8%), while hookworm (97.0%) was the predominant STH infection in Cambodia. In both populations, most STH infections were of light intensity (Table 18).

TABLE 18 CHARACTERISTICS OF THE STUDY POPULATIONS . Cuba ( N = 1389) Cambodia ( N = 2471) n (%) or mean ± sd n (%) or mean ± sd Age (years) 8.14 ± 2.07 9.68 ± 2.27 Sex (female) 640 (47.0%) 1236 (50.0%) Height for age z score 0.06 ± 1.04 −1.81 ± 1.05 Stunted 21 (1.6%) 1056 (42.9%) STH infection a 114 (8.4%) 302 (16.8%) Ascarislumbricoides 70 (5.2%) 5 (0.3%) Light (<5.000 epg) 55 (4.1%) 5 (0.3%) Moderate (5.000 –50.000 15 (1.1%) 0 epg) Heavy (>50.000 epg) 0 0 Trichuristrichiura 42 (3.1%) 6 (0.3%) Light (<1.000 epg) 38 (2.8%) 6 (0.3%) Moderate (1.000 –10.000 epg) 2 (0.1%) 0 Heavy (>10.000 epg) 2 (0.1%) 0 Hookworm 15 (1.1%) 293 (16.3%) Light (<2.000 epg) 13 (1.0%) 283 (15.8%) Moderate (2.000 –4.000 epg) 0 9 (0.5%) Heavy (>4.000 epg) 2 (0.1%) 1 (0.1%) Hair zinc (µg/g) 113 (91-137) b -- Zinc deficiency c 28 (12.2%) -- Plasma zinc d (µmol/L) -- 7.65 ± 1.69 Zinc deficiency e -- 1884 (92.8%) Inflammation No inflammation -- 1450 (60.5%) Only CRP elevated -- 8 (0.3%) Only AGP elevated -- 816 (34.1%) CRP & AGP elevated -- 122 (5.1%) a: N= 1353 (Cuba) or N = 1795 (Cambodia); b: median (IQR), N =230; c: hair zinc < 70 µg/g; d:N=2112; e: age 4 –9: plasma zinc < 9.9 µmol/L; girls age 10 and up: plasma zinc < 10.1 µmol/L boys age 10 and up: plasma zinc < 10.7 µmol/L, N = 2030.

Associations between Height for Age, Zinc and STH Infection

STH infected Cuban children had on average lower height for age compared to their uninfected peers (Table 19), and regression analysis showed a significant negative association between STH infection and height for age (Table 20). The association between hair zinc and height for age was not significant but did show a positive trend. In Cambodia, plasma zinc, but not STH infection, was significantly associated with height for age (Table 20). In both populations, STH x zinc interaction terms were not statistically significant. However, when stratifying for STH infection, in the uninfected Cuban children a significant, positive association (aB 0.471, p = 0.033) was found between hair zinc and height for age.

TABLE 19 ZINC AND HEIGHT FOR AGE IN STH INFECTED AND UNINFECTED CHILDREN .

N Zinc N Height for age z score concentration (mean ± sd) Cuba STH 160 112.55 (88.3 – 1251 0.11 ± 0.97 uninfected 136.0) a STH infected 70 113.35 (94.4 – 117 −0.31 ± 1.16 143.7) a Cambodia STH 1239 7.74 ± 1.70 b 1450 −1.81 ± 1.05 uninfected STH infected 254 7.52 ± 1.70 b 296 −1.84 ± 1.09 a. Hair zinc in µg/g, median (IQR); b. Plasma zinc in µmol/L, mean ± sd.;

TABLE 20 LINEAR REGRESSION MODELS OF HEIGHT FOR AGE BY STH INFECTION AND ZINC .

independent N aB a p variable Cuba b STH infection 226 −0.483 0.001 Zinc 0.335 0.082 Cambodia c STH infection 1448 −0.008 0.902 Zinc 0.033 0.029 a: regression coefficient; b: adjusted for sex and age in months; c: adjusted for sex, age in months and inflammation categories

In the Cuban study, the median hair zinc concentration was slightly higher in STH infected than in uninfected children (Table 19), but the result of the regression analysis was not statistically significant (Table 21). In contrast, STH infected children in the Cambodian study had on average lower plasma zinc concentrations than their uninfected peers (Table 20). This association was borderline significant (Table 21).

TABLE 21 LINEAR REGRESSION MODELS OF ZINC BY STH INFECTION . Variable N aB p value

Cuba a STH infection 230 0.068 0.206

Cambodia b STH infection 1795 −0.233 0.051

a: adjusted for sex, age categories and height for age categories; b: adjusted for inflammation categories, sex, age categories and height for age categories

Discussion

The present study showed different associations between height for age, STH infection and zinc in Cuban and Cambodian schoolchildren. In the Cuban study population STH infection was significantly associated with lower height for age, while hair zinc concentrations were not. Conversely, in the Cambodian study population plasma zinc, but not STH infection, was significantly associated with higher height for age.

The two populations were markedly different in mean height for age. The Cuban schoolchildren were on average taller than the reference population 157 and stunting was rare. These characteristics generally indicate an adequate zinc status at population level 203 and this was confirmed by the observed hair zinc values. STH infection appeared to have a stronger effect than zinc on height for age in Cuban children. Because stunting was rare in the Cuban study population, the associations occurred in children of normal height. The Cambodian schoolchildren included in the study had a low mean height for age compared to the reference population 157 and stunting was common. The observed stunting suggested a zinc deficient population 203 , which was indeed corroborated by the observed plasma zinc values. In these children, STH infection was not associated with height for age.

This study also examined the relation between zinc and STH infection. Plasma zinc concentrations were lower in STH infected Cambodian children than in their uninfected peers. This association was borderline significant. Few other studies have addressed associations between zinc and STH infection. In 2009, Rosado et al . found that while zinc supplementation increased height for age in Mexican infants, this effect was diminished by Ascaris infection 342 . Kongsbak et al . found T. trichiura to be a significant predictor of serum zinc in a Bangladeshi population where stunting was common 331 . In this study, T. trichiura had a larger effect on serum zinc than did A. lumbricoides , suggesting species-specific differences. Osei et al . did not find serum zinc to differ significantly between STH infected and uninfected Indian children 343 . Two recent meta-analyses found no significant effect of zinc supplementation on STH (re-) infection rate 344,345 . The present study did not distinguish between the effects of the different STH species. In our Cambodian study, children carried almost exclusively hookworm infections. Hence, STH species-specific effects on zinc could not be determined in this population. Likewise, a comparison between zinc deficient and zinc sufficient children in STH infection was not possible, since almost all of the Cambodian children were zinc deficient.

The different associations between STH and stunting found in the two populations might reflect the difference in predominating STH species. In the present study, the Cuban children were

93 more often infected with A. lumbricoides or T. trichiura , while hookworm was the prevailing STH infection in Cambodia. These species have distinct life cycles and might therefore have quite different effects on nutritional status 320,330 . Recently, in a study conducted in children in the Philippines, Papier et al . showed that the proportion of stunted children was significantly higher among children infected with hookworm than among children infected with A. lumbricoides , and T. trichiura 346 . These findings are corroborated by the results of this study.

This study has some limitations, warranting caution in its interpretation. Since the present study is cross-sectional, causality cannot be inferred. STH infections and zinc deficiency are often put forward as important causes of child stunting 203,330 . However, reduced height for age might also reflect a generally poor nutritional status, which can influence both zinc uptake and susceptibility to infections. Stunting is also strongly related to poverty, as are STH infections and zinc status 329,347 . Moreover, observed associations between height, zinc and STH might all be explained in the context of ‘environmental enteropathy’; repeated exposure to intestinal pathogens resulting in inflammation and remodeling of the mucosa, causing widespread malabsorption 348 .

Associations between zinc and helminths can also be interpreted in various ways. STH infection might damage or block the intestinal mucosa, resulting in reduced uptake of nutrients 330 . Additionally, the STH might compete with the host for essential elements. Inflammation resulting from infection can also lead to reduced micronutrient levels in plasma, induced by the acute phase response 349 . For this reason, inflammation was taken into account in the present analysis. On the other hand, zinc status can influence susceptibility to infection by its effects on immune function 203 .

While the importance of assessing zinc levels has been recognized for many years, a reliable and representative method to measure zinc remains a challenge. Serum or plasma zinc is considered the best available biomarker of zinc deficiency in populations 203 . It has been shown that plasma zinc reflects dietary zinc intake and that it responds consistently to zinc supplementation 203,350 . However, the timing of blood collection and fasting status influence the zinc concentrations measured in plasma 351 . Moreover, zinc is considered a ‘type -II’ nutrient, meaning that no real stores exist, and that growth faltering is one of the key features of deficiency 352 . Associations between low zinc concentration in hair and poor growth have been documented 203 . Hair zinc has been shown to increase after supplementation 351 . However, it has been argued that zinc in hair reflects a more extended period of exposure than plasma zinc 203 . It cannot be excluded that differences observed in the present study might be (partly) due to the use of different methods of zinc measurement. Presently, there are no reliable data on the correlation between hair zinc values and plasma or serum zinc values. Moreover, although the effects of the acute phase response on plasma zinc levels are widely recognized, there is currently no standard method of accounting for this in school-age children 339,353 .

Based on the results of this study, we recommend that STH infection and zinc status at population level should be taken into account when assessing the potential factors contributing to stunting. It is essential to define a standard and reliable method of measuring zinc and accounting for inflammation effects in order to further elucidate associations between zinc, STH infection and growth. In populations living in STH endemic areas, a possible association between zinc and STH should be considered. This will improve (the evidence base for) interventions on child growth, for instance by pairing zinc supplementation with helminth control strategies.

Chapter 4. Effectiveness of multi- micronutrient fortified rice through a school feeding program: Case study in Cambodia

4.1. Impact of multi-micronutrient fortified rice on hemoglobin, iron and vitamin A status of Cambodian school-aged children: a double-blind randomized controlled trial (Published paper)

Published in Nutrients (2016)

Marlène Perignon a, Marion Fiorentino a, Khov Kuong e, Marjoleine A Dijkhuizen n, Kurt Burja f, Megan Parker g, Chhoun Chamnan e, Jacques Berger a, Frank T Wieringa a

Introduction Micronutrient deficiencies, also known as hidden hunger, remain a critical public health problem affecting a third of the world’s population 296 . Iron deficiency (ID), the primary cause of anemia, has adverse effects on both human health and socio-economic development, with increased susceptibility to infections, elevated risk of maternal and child mortality, impaired cognitive and physical development of children and lower work productivity of adults 208,354 . Like ID, vitamin A deficiency (VAD) ranks among the 15 leading causes of the global burden of disease and was estimated to be responsible for 0·6 million deaths in children under 5 y of age 300 . VAD can cause xerophthalmia and impairs the immune system thereby increasing the severity and mortality risk of infectious diseases such as measles and diarrheal disease 85 . The 2011 estimates suggest anemia affects around 800 million children and women worldwide 354 . Anemia is primarily caused by iron deficiency but also by other micronutrient deficiencies such as vitamins B2, folate, and B12. Also vitamin A, selenium and copper have been associated with anemia 208 .Non-nutritional causes of anemia include acute and chronic diseases like malaria, HIV, and tuberculosis, or heavy blood loss such as that associated with intestinal parasite infections 208 . Hemoglobinopathies, one of the most common human genetic disorders 355 , must also be considered a factor of anemia, especially in South-East Asia where thalassemias are common 356 . Both women of reproductive age and children are the populations most at risk for anemia and micronutrient deficiencies. Approximately 273 million of children (43%), 32million pregnant women (38%), and 496million non-pregnant women (29%) were estimated to be anemic in 2011 354 . In Cambodia, undernutrition remains a major problem as large segments of the child population (6-59 months) are affected by stunting (40%), wasting (11% ), and anemia (55%) 295 . Micronutrient deficiencies and malnutrition are also widely spread in school-children: it is estimated that iron and vitamin A deficiencies affect 20% of school-aged children in South-East Asia, while 30% are zinc or iodine deficient 17 . Micronutrient deficiencies during school years can impair physical and mental development and reduces school attendance by increasing morbidity. Some studies reported that it is still possible to improve cognition at school age by improving micronutrient status 53,68,357 as well as positive effects on morbidity and growth, but the overall effects on these outcomes were equivocal and more evidence is required from studies in different contexts. The inclusion of micronutrient rich foods in the daily diet, like meat and a variety of vegetables and fruits, is often not affordable for populations living under conditions of poverty in both developed and developing countries. Food fortification is a cost-effective alternative to food-based approaches for controlling and preventing micronutrient deficiencies, and could improve the nutritional status of populations at risk. The Copenhagen Consensus 2008 actually ranked micronutrient fortification among the top three international development priorities using a cost-benefit analysis 358 . The fortification of staple foods is advantageous because it does not require the target population to change their dietary habits and allows fortification with multiple micronutrients since deficiencies often occur concurrently 69 . Many studies carried out in Latin America, Africa and India showed that rice fortification is safe and effective in improving micronutrient status, with impact depending on the micronutrient content of the fortified rice 130,176,359,360 . In rice consuming countries such as Cambodia, multi-micronutrient fortified rice 96 could be a promising strategy to address micronutrient deficiencies. However, evidence of impact is needed by the Cambodian government and WFP to support including fortified rice in food-based social safety net programmes or as a potential vehicle in the government’s proposed national food fortification guidelines. Consequently, the objective of the FORISCA UltraRice+NutriRice study, a large scale cluster-randomized, double-blinded, placebo-controlled trial, was to evaluate the impact of three different types of multi- micronutrient fortified rice distributed through the WFP school-meal program (SMP) on micronutrient status, health, and cognition of Cambodian schoolchildren. This paper examines the impact of fortified rice on hemoglobin, iron and vitamin A status.

Subjects and Methods

Study site The study was conducted between November 2012 and July 2013 in 20 primary schools from 5 districts of Kampong Speu province in Cambodia. Kampong Speu is one of Cambodia’s 23 provinces, situated 60 km west of Phnom Penh, the capital city. Agriculture is predominant, with rice farming as the main occupation and income source.

Study design A total of 20 primary schools were selected to constitute four intervention groups (including placebo) and a control group. Sixteen schools (intervention groups and placebo) were selected from the primary schools participating in the World Food Programme (WFP) school meal program. This program provides children with a daily breakfast consisting of rice, beans, canned fish, iodine-fortified salt, and vegetable oil enriched with vitamins A and D. The 16 selected schools were randomly allocated to 1 of the 4 intervention groups using a computer generated list with predefined criteria of group size. Randomization was done by one of the researchers (MAD) not involved in the field work and the codes were not known by any researchers or field staff during implementation, thus assuring the study was double-blinded. The four intervention groups were: 1) fortified cold-extruded rice UltraRice original formulation (URO), 2) fortified hot-extruded rice UltraRice new formulation (URN), 3) fortified hot- extruded rice Nutririce and 4) non-fortified rice (placebo). Four schools were randomly selected from sixteen primary schools participating in another program of WFP (take-home ration program) but not receiving a school meal so as to constitute the control group to assess the impact of the normal school meal program, and the additional benefits of including fortified rice. Prior to the study, all parents of children from the 20 schools were invited to attend a meeting at which the study procedures were explained. Written informed consent was obtained from the parents as was verbal assent from the participating children. Children attending the selected schools were eligible to be part of the study if they were 6-16 y of age, written informed consent was obtained from parent/caregiver, and the child did not have a mental or severe physical handicap. Children with severe anemia (defined as hemoglobin concentration<70 g/L) were excluded, but received multiple micronutrient supplements for 2 months, after which hemoglobin concentrations were re-assessed. A sample size of 500 children per group was calculated to detect a difference in Hb concentration of 4 g/L, assuming an average Hb concentration of 110g/L 361 . Other outcomes such as changes in FER, TfR and RBP concentrations all needed smaller sample sizes. In each school, 132 children were randomly selected after stratification by sex and grade, hence 528 children per group and a total of 2640 children. Two hundred children were not recruited because of absence on the day of data collection or refusal of participation (n=90), age outside age criteria (n=107) or severe anemia (n=3). A total of 2440 schoolchildren aged 6-16 y participated in the study. Figure 6 shows the subject selection scheme of the study.

16 schools participating in THR program of WFP and 16 schools participating in WFP SMP randomly allocated to 4 with >400 intervention groups children (n=8279)

random selection of 4 schools

Control Placebo URO URN Nutririce n=2034 n=1665 n=1500 n=1500 n=1580

Pre-selected by Control Placebo URO URN Nutririce randomization (n=2640) n=528 n=528 n=528 n=528 n=528

exclusion due to age (n=107) n=17 n=24 n=26 n=15 n=25 exclusion due to severe anemia (n=3) n=1 n=1 n=0 n=0 n=1 exclusion due to absence or refuse (n=90) n=20 n=24 n=26 n=17 n=3

Eligible (n=2440) Control Placebo URO URN Nutririce n=490 479 n=476 n=496 n=499

Follow-up at midline Control Placebo URO URN Nutririce (n=2107) n=415 n=430 n=432 n=435 n=395

dropped school (n=56) n=7 n=12 n=12 n=11 n=14 transferred to another school (n=15) n=3 n=3 n=0 n=2 n=7 refused to participate (n=27) n=8 n=12 n=3 n=2 n=2 severe anemia at midline (n=1) n=0 n=1 n=0 n=0 n=0 absence (n=93) n=21 n=24 n=15 n=17 n=16 Follow up at endline Control Placebo URO URN Nutririce (n=2248) n=451 427 n=446 n=464 n=460

FIGURE 6 TRIAL PROFILE . THR: TAKE -HOME RATION , URO: ULTRA RICE ORIGINAL FORMULATION , URN: ULTRA RICE NEW FORMULATION , WFP-SMP: WORLD FOOD PROGRAMME SCHOOL -MEAL PROGRAM

The primary outcomes evaluated in the FORISCA UltraRice+NutriRice study were the prevalence of anemia and micronutrient deficiencies, anthropometry, health and general well-being, and cognitive function. Prevalence of helminth infection, gut flora, and immune function were evaluated as secondary outcomes. This paper focuses on the impact of the intervention on the prevalence of anemia, evaluated using hemoglobin concentration, and iron and vitamin A deficiencies, respectively evaluated using FER and TfR and RBP plasma concentrations. The study was approved by the National Ethic Committee for Health Research (NECHR) of the Ministry of Health, Phnom Penh, Cambodia, the Ministry of Education, Youth and Sports, Phnom Penh, Cambodia, and the Research Ethical committee (REC) of PATH, Seattle, USA. Written informed consent was collected from all parents/caregivers of children prior to the enrolment in the study.

Intervention The standard WFP-SMP breakfast consists of 115 g of (uncooked) rice, 15 g of canned fish, 15 g of yellow split peas, 5 g of oil (fortified with vitamin A and vitamin D) and 3 g of salt (iodized). Breakfast was distributed 6 days/week during 6 months. Fortified ‘kernels’, produced by extrusion, were provided by PATH (UltraRice technology) and DSM (Nutririce). The UltraRice original (URO) was produced by cold extrusion and the UltraRice new (URN) and Nutririce by hot extrusion. The fortified rice for consumption was then obtained by blending the kernels at a ratio of 1/100 with the same local unfortified rice used for placebo group. Blending was done under supervision of WFP at a local food factory in Phnom Penh. Rice

98 was packaged in bags containing a letter (A – H) according to allocation to intervention group, with 2 letters per intervention group to strengthen blinding. The micronutrient contents of the three different types of fortified rice (URO, URN and Nutririce) are given in Table 22. A previous study conducted in primary schools located in the same region in Cambodia showed good acceptability of fortified rice by parents and children 179 .Participant micronutrient status was evaluated at baseline and after 3 and 6 months of the intervention. Children were dewormed using mebendazole just after baseline and endline, according to the standard procedures of the Ministry of Health, Cambodia.

TABLE 22 MICRONUTRIENT CONTENTS OF THE FORTIFIED RICES PER 100 G OF UNCOOKED BLENDED RICE . URO: ULTRA RICE ORIGINAL FORMULATION , URN: ULTRA RICE NEW FORMULATION .

Micronutrients URO URN NutriRice Iron(mg) 10.67 7.55 7.46 Zinc (mg) 3.04 2.02 3.68 Vitamin B1 (mg) 1.06 1.43 0.69 Folic acid (mg) 0.17 0.28 0.14 Vitamin A (IU) - 2140 960 Vitamin B3 (mg) - 12.57 7.98 Vitamin B12 ( mg) - 3.8 1.26 Vitamin B6 (mg) - - 0.92

Blood samples collection Blood samples (5ml) were collected by venipuncture and aliquoted into trace-element free vacutainers with no anticoagulant (Vacuette, Greiner Bio One) and into EDTA tubes (2ml) for hemoglobinopathy analysis. Samples were then stored in a cool box containing ice-packs and transported to Phnom Penh within 5 h of blood collection. The blood samples were centrifuged at 2700 rpm (1300g) for 10 min at room temperature. Serum was aliquoted in capped Eppendorf tubes and stored at -30°C until transfer for analysis. The anticoagulated blood samples were transported to the Institut Pasteur du Cambodge for hemoglobinopathies analysis by electrophoresis (MINICAP System).

Hemoglobin concentration Hemoglobin concentrations were measured immediately after blood taking using the HemoCue (301+ system, HemoCue Angholm, Sweden). The HemoCue system was controlled on each day of blood collection using three levels of blood controls (HemoTrol®). Anemia was defined as hemoglobin concentration <115 g/L for children <12 y, <120 g/L for children between 12 and 15 y and girls ³15 y, and <130 g/L for boys ³ 15 y according to WHO guidelines 308 .

Blood samples analysis: markers of iron, vitamin A and inflammation status Serum samples were transported on dry ice to the VitMin laboratory (Willstaett, Germany) for determination of retinol-binding protein (RBP), ferritin (FER), soluble transferrin receptors (TfR), C- reactive protein (CRP), and α1 -acid-glycoprotein (AGP) concentrations. RBP, FER, TfR, CRP, AGP were measured by a sandwich enzyme-linked immunosorbent assay (ELISA) technique 195 . Inflammation was defined as high CRP (>5mg/L) and/or high AGP concentrations (>1g/L), and categorized in four groups based on inflammation markers levels: no inflammation (normal CRP and AGP), incubation phase (high CRP and normal AGP), early convalescence phase (both CRP and AGP elevated), and late convalescence phase (high AGP only) 197 . Serum FER levels can be affected by infection or inflammation, therefore FER concentration was adjusted by multiplying values by correction factors published by Thurnham et al. , namely 0.77, 0.53 and 0.75 for children respectively in incubation, early convalescence, and late convalescence phases (19). While ferritin is a positive acute phase protein that is elevated in the presence 99 of inflammation, TfR is not significantly affected by infection or inflammatory processes 308 . Depleted iron stores were defined by low FER (corrected value<15 µg/L) 308 , and tissue iron deficiency by high TfR (>8.3mg/L) 210,362 . Iron deficiency was defined by depleted iron stores and/or tissue iron deficiency. Total body iron (BI) was calculated from FER corrected for inflammation and TfR as described by Cook et al. 199 . A cut-off of 4mg/kg of body weight was used to define marginal body iron stores. Serum retinol is bound with RBP in a 1-to-1 complex 200 , hence RBP concentrations were used as a proxy for more conventional circulating retinol concentrations to evaluate vitamin A status. RBP concentrations were adjusted for the presence of inflammation by multiplying values by correction factors of 1.15, 1.32 and 1.12 for incubation, early convalescence and late convalescence phases respectively 201 . Vitamin A deficiency (VAD) was defined by corrected RBP < 0.70 μmol/L, severe vitamin A deficiency by corrected RBP <0.35 μmol/L and marginal vitamin A status by corrected RBP values ³0.7 and < 1.05 μmol/L 200,310 .

Parasite infestation On the day of data collection, children received a plastic container and instructions for fecal sample collection and were requested to return a fecal sample to the school the following day. Samples were then stored at 4°C and analyzed by the National Malaria Center (CNM, Phnom Penh, Cambodia) using the Kato- Katz method 314 . The parasite egg output was recorded as eggs per gram feces (epg).

Data management and Statistical analysis Data entry and validation by double entry of questionnaires was performed using EpiData version 3.1 software (EpiData Association, Odense, Denmark). Data management and analyses were performed using SPSS version 20.0 software (SPSS, Inc., Chicago, IL). Normality of distributions was evaluated using Kolmogorov-Smirnoff test. Not normally distributed data were considered to be reasonably close to normality to allow parametric tests when skewness and kurtosis values ranged between -1.0 and +1.0 363 . Continuous variables that were not normally distributed were log-transformed. Baseline characteristics were compared between intervention groups using ANOVA and Pearson’s chi -square tests. Risk factors for anemia were analysed using binary logistic regression. Generalized mixed models (linear or binary logistic regression) were used to evaluate the effects of time, group and time x group interaction on Hb, FER, TfR and prevalence of VAD, while taking into account random effects of individuals and school clusters. Primary analysis were performed including age, gender, hemoglobinopathy and baseline characteristics (inflammation, parasite infection, Hb, iron and VA status) in the model. Final models were adjusted on variables identified as having a significant fixed effect in the primary analysis. Multiple comparisons were conducted by using the Bonferroni post-hoc test. The significance level was set at 5 % (p < 0.05) for all tests.

Results

Of the 2440 children included at baseline, 192 did not complete the study (control: n=39, placebo: n=52, URO: n=30, URN: n=32, Nutririce: n=39) due to absence on the day of data collection (n=93), dropping out of school (n=56), transfer to another school (n=15) or refusal to cooperate (n=27) (Figure 6). One child received treatment for severe anemia at midline and was consequently excluded from follow-up. Baseline characteristics of participants in each group are presented in Table 23.There were no significant differences in age and gender between the study groups. The mean ± SD age of children at baseline was 9.6 ±2.3 y and half (49.9%) were girls. Placebo and MMFR groups did not differ in baseline characteristics for Hb, body iron, prevalence of anemia, ID, and iron deficiency anemia (IDA). However, despite the randomization, the prevalence of marginal VA status, FER and TfR levels, inflammation, parasite infection and hemoglobinopathy significantly differed between groups. Furthermore, the control group was significantly different from the placebo and intervention groups for most indicators. This difference was

100 expected as schools in the control group were not selected to be part of the WFP school meal program precisely because of their better status according to poverty, food insecurity and education indicators. Inflammation (CRP>5 mg/L and/or AGP>1 g/L) was found in more than one third of the children (39.5%) and parasite infection in 18% of the children. Only 1.4% of the children had depleted iron stores (FER<15 μg/L) whereas 51% had tissue iron deficiency (TfR>8.3 mg/L). Only 2% of the childre n had negative body iron stores but marginal body iron stores (total body iron<4 mg/kg body weight) was more prevalent (13.9%). Prevalence of iron deficiency, as defined by low FER and/or high TfR, was 51.2%, including 10% of iron-deficiency anemia. Prevalence of vitamin A deficiency (VAD) was 0.7%, with no severe VAD, whereas 7.9% of the children had marginal VA status (0.7≤corrected RBP<1.05 μmol/L). Using a higher cut-off of 0.725μmol/Las suggested by Hix et al. 364 led to higher though similar prevalence: 1.2% of VAD and 10.2% of marginal VA status. At baseline, the prevalence of anemia in all schools was 15.6%. Anemia was multi-factorial with hemoglobinopathy, VAD and depleted iron stores being the strongest risk factors (all p<0.01) ( Table 24 ).

The intervention had a significant effect on Hb and iron status when compared with the placebo group (interaction effect: p<0.001 for all) ( Table 25 ). After 3 months, Hb significantly increased by0.8 g/L for children receiving URN rice when compared with children receiving unfortified rice (p=0.048), but at the end of the intervention, no significant differences remained between the groups. The FER concentration significantly increased by 8 and 10 mg/L in Nutririce and URN groups after 6 months of the intervention (p<0.001). TfR concentrations also increased in those two groups, after 3 (p<0.05) and 6 months (p<0.001). No significant difference was found for the group receiving URO, although Hb and TfR tended to decrease in the first 3 months. The intervention had no effect on total body iron.

Inflammation status functioned as a significant effect modifier of the intervention on Hb and iron status (Table 26 ). For children with no inflammation (both CRP<5mg/L and AGP <1g/L) at baseline, midline and endline, Hb concentration significantly increased by 2.1 g/L after 3 months in URN group when compared to the placebo group (p<0.01). The increase was still significant after 6 months in this group (+1.8 g/L, p=0.015). Although not statistically significant, Hb also tended to increase after 6 months in the two other groups receiving fortified rice, URO and Nutririce (p=0.054 and p=0.095, respectively) for this sub-sample of children with no inflammation. TfR concentrations were significantly increased after 6 months in both URN and Nutririce groups (p<0.001). Increase of FER was significant in URN group when compared with placebo (p<0.001), and there was a trend for higher FER in the children receiving Nutririce (p=0.07). No significant difference was found in prevalence of anemia.

The intervention had a significant impact on vitamin A status, with a lower prevalence of marginal vitamin A status in children receiving fortified rice including vitamin A i.e. URN and Nutririce ( Table 27 ). After 6 months, these children had respectively 4 times (OR=0.24, p<0.001) and 5 times (OR=0.20, p<0.001) less risk of marginal VA status than children in the placebo group. Risk was reduced by almost 50% (OR=0.52, p<0.05) after 3 months for children in Nutririce group.

101

TABLE 23 BASELINE CHARACTERISTICS OF ALL CHILDREN PARTICIPATING IN THE STUDY AND FOR EACH INTERVENTION GROUP

p- ALL CONTROL PLACEBO URO URN NUTRIRICE value 5

N 2440 490 479 476 496 499

Age (y) 9.65 ± 2.26 9.82 ± 2.30 9.61 ± 2.28 9.55 ± 2.14 9.64 ± 2.22 9.60 ± 2.35 NS

% girls 49.9 (n=1218) 50.6 (n=248) 49.9 (n=239) 49.6 (n=236) 50.6 (n=251) 48.9 (n=244) NS

% inflammation 39.5 (n=935) 45.6 a (n=215) 43.5 a (n=201) 42.3 a,b (n=199) 32.3 c (n=158) 34.0 b,c (n=162) <0.05

% parasite 17.9 (n=323) 9.4 a (n=33) 22.9 b (n=83) 23.9 b (n=89) 19.5 b,c (n=74) 12.9 a,c (n=44) <0.05 infection

Hemoglobinopathy (% HbE)

Hb E ≤5% 54.1 (n=1130) 57.4 a,b (n=236) 60.1 b (n=256) 52.7 a,b (n=225) 50.7 a,b (n=218) 49.5 a (n=195) <0.05

Hb E 5-80% 37.8 (n=789) 36.0 a,b (n=148) 31.7 b (n=135) 41.2 a (n=176) 39.5 a,b (n=170) 40.6 a,b (n=160) <0.05

HbE>80% 8.1 (n=169) 6.6 (n=27) 8.2 (n=35) 6.1 (n=26) 9.8 (n=42) 9.9 (n=39) NS

ANEMIA and IRON STATUS

Hb (g/L) 124.2 ± 0.2 125.6 ± 0.4 a 123.6 ± 0.4 b 124.3 ± 0.4 a,b 123.6 ± 0.4 b 124.1 ± 0.4 a,b <0.05

% anemia 15.6 (n=376) 9.8 a (n=48) 18.9 b (n=89) 15.3 a,b (n=72) 17.8 b (n=88) 16.4 b (n=79) <0.05

76.2 ± 36.9 83.0 ± 35.8 a 77.0 ± 36.5 a,b 79.7 ±38.8 a 69.7 ± 36.3 c 71.9 ± 35.3 b,c FER 1 (mg/L) <0.05 (n=2368) (n=471) (n=462) (n=470) (n=489) (n=476)

% FER 1 <15 mg/L 1.4 (n=34) 0.2 a (n=1) 0.4 a (n=2) 0.9 a (n=4) 3.7 b (n=18) 1.9 a,b (n=9) <0.05

8.8 ± 2.5 8.5 ± 2.1 a 9.0 ± 2.4 b 9.1 ± 2.5 b 8.4 ± 2.5 a 8.9 ± 3.0 a,b TFR (mg/L) <0.05 (n=2368) (n=471) (n=462) (n=470) (n=489) (n=476)

% TFR > 8.3 51.0 (n=1207) 46.7 a (n=220) 55.6 a,b (n=257) 56.6 b (n=266) 47.0 a (n=230) 49.2 a,b (n=234) <0.05 mg/L

% ID 2 total 51.2 (n=1212) 46.7 a (n=220) 55.6 a,b (n=257) 56.6 b (n=266) 47.6 a,b (n=233) 49.6 a,b (n=236) <0.05

% ID 2 with 9.9 (n=235) 6.4 a (n=30) 12.1 b (n=56) 10.4 a,b (n=49) 10.6 a,b (n=52) 10.1 a,b (n=48) <0.05 anemia

Body iron 6.0 ± 2.2 6.5 ± 1.8 a 6.0 ± 2.0 b 6.0 ± 2.2 b 5.8 ± 2.4 b 5.8 ± 2.4 b <0.05 (mg/kg) (n=2368) (n=471) (n=462) (n=470) (n=489) (n=476)

% BI ≥ 4 mg/kg 86.2 (n=2042) 92.6% a (n=436) 87.2 a,b (n=403) 86.2 b (n=405) 83.8 b (n=410) 81.5 b (n=388) <0.05

% BI 0-4 mg/kg 11.9 (n=281) 6.8% a (n=32) 11.7 a,b (n=54) 12.6 b (n=59) 12.7 b (n=62) 15.5 b (n=74) <0.05

% BI < 0 mg/kg 1.9 (n=45) 0.6 a (n=3) 1.1 a,b (n=5) 1.3 a,b (n=6) 3.5 b (n=17) 2.9 a,b (n=14) <0.05

102

VITAMIN A STATUS

1.58 ± 0.43 1.60 ± 0.39 a,d 1.62 ± 0.43 a, b 1.69 ± 0.43 b 1.48 ± 0.43 c 1.52 ± 0.44 c,d RBP 1 (mmol/L) <0.05 (n=2368) (n=471) (n=462) (n=470) (n=489) (n=476)

% marginal VA 7.9 (n=188) 5.3 a (n=25) 6.9 a,b (n=32) 3.2 a (n=15) 12.9 c (n=63) 11.1 b,c (n=53) <0.05 status 3

% VAD 4 0.7 (n=17) 0.2 (n=1) 0.4 (n=2) 0.2 (n=1) 1.2 (n=6) 1.5 (n=7)

Results are mean ± SD unless stated, 1corrected for inflammation using multiplier correction factors published by Thurnham et al. (19,24), 2based on FER 1 < 15 mg/L and/or TfR> 8.3 mg/L, 30.7 ≤ RBP 1< 1.05 µmol/L, 4RBP 1< 0.7 µmol/L, 5from ANOVA test. Groups in the same subset (a, b or c) do not differ significantly from each other’s at the 0.05 level (Bonferroni post-hoc test).BI: body iron, Hb: hemoglobin, FER: ferritin, TfR: transferrin receptors, ID: iron-deficiency, NS: not significant, URO: UltraRice original formula, URN: UltraRice new formula, RBP: retinol binding protein, VA: vitamin A, VAD: vitamin A deficiency

103

TABLE 24 RISK FACTORS FOR ANEMIA AT BASELINE

Adjusted OR 2 Factors p-value (95% CI)

Gender (reference: male) 0.86 (0.62; 1.19) 0.355

Age 1.03 (0.96; 1.12) 0.378

Parasite infection (reference: no infection) 1.63 (1.10; 2.42) 0.016

Inflammation (reference: no inflammation)

Incubation 1.95 (0.26; 14.58) 0.514

early convalescence 2.20 (1.08; 4.48) 0.029

late convalescence 1.18 (0.82; 1.69) 0.375

Hemoglobinopathy (reference: HbE< 5%)

5% ≤ HbE< 80% 1.87 (1.30; 2.69) 0.001

HbE ≥ 80% 24.10 (15.09; 38.49) <0.001

VA status (reference: normal VA status)

marginal VA status (0.7 < RBP 1< 1.05 mmol/L) 1.57 (0.91; 2.72) 0.106

VAD (RBP 1< 0.7 mmol/L) 8.56 (2.30; 31.89) 0.001

Depleted iron stores (FER 1< 15 mg/L) 52.97 (11.43; 245.55) <0.001

Tissue iron deficiency (TfR> 8.3 mg/L) 0.99 (0.70; 1.42) 0.979

1corrected for inflammation, 2from binary logistic regression, adjusted for age, gender, parasite infection, inflammation, hemoglobinopathy, VA and iron status. Hb: hemoglobin, FER: ferritin, TfR: transferrin receptors, VA: vitamin A, VAD: vitamin A deficiency.

104

TABLE 25 BIOCHEMICAL OUTCOMES AND EFFECT SIZES AFTER 3 AND 6 MONTHS OF INTERVENTION FOR ALL PARTICIPATING CHILDREN

Hb (g/L) FER 1 (mg/L)

interaction term 2 interaction term 2 n Mean SE n mean SE β coefficient (95% CI) p-value β coefficient (95% CI) p-value

Placebo 470 123.7 1.3 - 462 77.5 3.5 - URO 471 124.7 1.3 - 470 79.9 3.5 - B URN 494 123.7 1.3 - 489 70.0 3.5 -

Nutririce 482 124.4 1.3 - 476 72.1 3.5 - Placebo 428 123.3 1.3 - 426 69.0 3.5 - URO 428 123.5 1.3 -0.74 (-1.54; 0.06) 0.068 428 68.4 3.6 -3.08 (-7.22; 1.06) 0.144 M URN 434 124.1 1.3 0.80 (0.01; 1.59) 0.048 347 64.2 3.6 2.72 (-1.57; 7.02) 0.214 Nutririce 394 124.4 1.3 0.50 (-0.31; 1.31) 0.230 393 66.5 3.6 2.88 (-1.33; 7.09) 0.180 Placebo 425 122.6 1.3 - 421 71.6 3.6 - URO 445 124.1 1.3 0.51 (-0.28; 1.30) 0.207 443 72.6 3.5 -1.46 (-5.58; 2.65) 0.486 E URN 464 123.0 1.3 0.36 (-0.42; 1.14) 0.368 463 74.8 3.5 10.70 (6.62; 14.78) <0.001 Nutririce 454 123.5 1.3 0.19 (-0.60; 0.98) 0.633 450 74.5 3.5 8.32 (4.19; 12.44) <0.001 TFR (mg/L) Body iron (mg/kg)

interaction term 2 interaction term 2 n mean SE n mean SE β coefficient (95% CI) p-value β coefficient (95% CI) p-value Placebo 462 8.98 0.24 - 462 6.01 0.27 - URO 470 9.11 0.24 - 470 6.05 0.27 - B URN 489 8.42 0.24 - 489 5.79 0.27 - Nutririce 476 8.87 0.24 - 476 5.78 0.27 - Placebo 426 8.11 0.24 - 426 5.99 0.27 - URO 428 7.98 0.24 -0.26 (-0.53; 0.01) 0.059 428 5.94 0.27 -0.09 (-0.30; 0.13) 0.427 M URN 347 7.90 0.24 0.34 (0.06; 0.62) 0.017 347 5.75 0.27 -0.01 (-0.23; 0.21) 0.928 Nutririce 393 8.49 0.24 0.49 (0.21; 0.76) 0.001 393 5.62 0.27 -0.13 (-0.35; 0.08) 0.227 Placebo 421 8.18 0.24 - 421 6.11 0.27 - URO 443 8.08 0.24 -0.24 (-0.50; 0.04) 0.088 443 6.09 0.27 -0.06 (-0.27; 0.15) 0.582 E URN 463 8.51 0.24 0.89 (0.62; 1.15) <0.001 463 6.00 0.27 0.11 (-0.09; 0.32) 0.284 Nutririce 450 8.74 0.24 0.66 (0.39; 0.93) <0.001 450 5.87 0.27 -0.01 (-0.22; 0.20) 0.947

Results are mean ± SE unless stated, 1corrected for inflammation, 2Generalized linear mixed models adjusted for age, gender and baseline characteristics were used to evaluate the effects of time x group interaction term, B: baseline, M: midline, E: endline, Hb: hemoglobin, FER: ferritin, TfR: transferrin receptors, URO: UltraRice original formula, URN: UltraRice new formula.

105

TABLE 26 BIOCHEMICAL OUTCOMES AND EFFECT SIZES AFTER 3 AND 6 MONTHS OF INTERVENTION FOR THE SUB -SAMPLE OF CHILDREN WITH NO INFLAMMATION AT BASELINE , MIDLINE AND ENDLINE

Results are mean ± SE unless stated, 2Generalized linear mixed models adjusted for age, gender and baseline characteristics were used to evaluate the effects of time x group interaction term, B: baseline, M: midline, E: endline, Hb: hemoglobin, FER: ferritin, TfR: transferrin receptors, URO: UltraRice original formula, URN: UltraRice new formula.

106

TABLE 27 PREVALENCE OF MARGINAL VA STATUS AFTER 3 AND 6 MONTHS OF INTERVENTION AMONG ALL CHILDREN .

All children

interaction term

Adjusted OR 1 p- n % (95% CI) (95% CI) value

Placebo 462 5.4 (2.9; 9.9) -

URO 470 2.6 (1.3; 5.4) - B URN 489 12.3 (7.2; 20.3) -

Nutririce 476 11.0 (6.4; 18.4) -

Placebo 426 11.0 (6.3; 18.6) -

2.55 (1.22; 428 URO 13.0 (7.6; 21.4) 5.33) 0.012

M 0.60 (0.33; 347 URN 15.4 (9.0; 25.1) 1.10) 0.101

0.52 (0.28; 393 Nutririce 12.2 (7.0; 20.4) 0.96) 0.036

Placebo 421 12.4 (7.1; 20.6) -

1.37 (0.65; 443 URO 8.3 (4.7; 14.5) 2.91) 0.410

E 0.20 (0.10; 463 URN 6.3 (3.4; 11.4) 0.37) <0.001

0.24 (0.13; 450 Nutririce 6.8 (3.7; 12.1) 0.45) <0.001

1mixed logistic regression model adjusted for age, gender and baseline characteristics was used to evaluate the effect of time x group interaction term. B: baseline, M: midline, E: endline, Hb: hemoglobin, FER: ferritin, TfR: transferrin receptors, URO: UltraRice original formula, URN: UltraRice new formula.

107

Discussion

This study is the largest to date to test the effectiveness of three types of MMFR to improve micronutrient status and reduce deficiencies among schoolchildren. Over the intervention period, consumption of fortified rice had a significant effect on iron and VA status when compared with the placebo group receiving normal rice. However, there was no overall impact after 6 months on hemoglobin concentrations, with Hb concentrations only 0.2 – 0.5 g/L higher in the fortified rice groups as compared to placebo. There was no impact on anemia prevalence either, which according to WHO classification, represented only a mild public health problem (15.7%) in this population 208 . The lack of impact on anemia prevalence might be explained by the multifactorial nature of anemia which was associated with low FER and VAD, but also with non-nutritional factors like parasite infection, inflammation and hemoglobinopathy. Hence, several factors could underlie this lack of impact on hemoglobin concentrations. First, the high prevalence of hemoglobinopathies in the study population may have blunted the effect of fortified rice on Hb concentrations, as it has been reported that women with thalassemia had a reduced iron incorporation after iron supplementation 365 . However, in our study, there was no difference between children with normal hemoglobin and hemoglobinopathies in terms of response to the intervention in hemoglobin concentrations, perhaps because the majority of the hemoglobinopathies in the present study consisted of HbE. Second, inflammation increases hepcidin concentrations, which reduces iron absorption from the gut 366 . Indeed, in our study population, the prevalence of sub-clinical inflammation was high and a significant effect modifier. In children without inflammation, all the 3 types of MMFR increased or tended to increase Hb concentrations, whereas there was no impact of fortified rice on hemoglobin concentrations in children with inflammation. The increase in hemoglobin concentrations in children without inflammation over the 6 mo intervention (1.2 – 1.8 g/L) was however small considering that 6 to 10 mg of iron was provided (dependent on MMFR), 6 days per week for 6 months. Third, the form of the iron used, ferric pyrophosphate (FePP), is known to have a lower bioavailability than ferrous sulfate 367 , but FePP is preferred however because of its superior organoleptic qualities (i.e. color, taste, smell). However, other studies using rice fortified with only FePP (and not other micronutrients) have significantly improved hemoglobin concentrations 359,360 . Finally, iron status at baseline might have been an important factor in the overall response in hemoglobin concentrations to the intervention. Actually, in the present study, the baseline prevalence of depleted iron stores, as reflected by FER concentration, was very low(1.4%) whether it was estimated using a cut-off of 15 μg/L with FER values corrected for inflammation (98.6%), or a higher cut-off of 30 μg/L with uncorrected values (95%) 308 . Surprisingly and in contrast to FER, >50% of the children had high TfR concentrations suggesting functional iron deficiency. Iron-deficient erythropoiesis (IDE) is actually the most common cause of elevated TfR 199 . The sequential process of development of iron deficiency generally starts with depletion of iron stores (low FER) leading to a lack of iron from the tissue (high TfR), IDE and finally to IDA. Although inconsistent with this general pattern of iron status biomarkers, the observed discrepancy between FER and TfR levels has already been reported in malaria and non-malaria environments 225,368-370 . Functional tissue iron deficiency can also occur despite normal or even increased storage iron, due to impaired release of iron from stores or impaired physiological systems for transporting iron to target tissues 198,315 . Moreover, TfR concentration depends both on the number of TfR per cell, a function of the iron status of the cell, and on the number of erythroid precursors in the bone marrow 371 . Thus, TfR reflects the

108 tissue iron needs but also the intensity of erythropoiesis. Some diseases common in developing countries, including thalassemia, megaloblastic anemia due to folate deficiency, or hemolysis due to malaria, may increase erythropoiesis and TfR independently of iron status 321 . Malaria may not be considered as a significant cause in our study since prevalence is very low in the study area. Hemoglobinopathy, on the other hand, was highly prevalent (45.9% children with abnormal Hb types>5%) and could thus be a potential explanation for the high TfR. Indeed, in our study population, TfR concentrations were significantly higher for children with >80% of abnormal Hb type, mainly HbE, than children with a normal Hb profile (+2.04 mg/L, 95%CI: 1.63; 2.45, p<0.001). However, there was no difference in TfR concentrations between children with normal Hb type (HbA>95%) and HbA levels between 20% and 95%, indicating that for example in children with heterozygote HbE, TfR concentrations were not significantly increased. In addition, in children with normal Hb, TfR was increased in 49% of the children, showing that in children without hemoglobinopathy, there was also a major discrepancy between iron stores and tissue iron needs, and thus indicating other causes for the elevated TfR. This discrepancy could also have been caused by sub-clinical inflammation which was highly prevalent (39.5%). Cytokines released during inflammation induce the production of hepcidin 198 , which then inhibits macrophages iron release and intestinal iron absorption 366 . This hypothesis is supported by the bigger effect of fortified rice on Hb concentration observed for children without any inflammation over the intervention period. However, simultaneous high FER and high TfR concentrations was also prevalent in the children without inflammation: in this sub-sample (n=1434), 42% of the children had an elevated TfR. TfR is thought to be less affected by inflammation than FER (38), but in the present study, children with inflammation had TfR levels significantly higher than children without inflammation (+1.15 mg/L, 95CI: 0.94; 1.37, p<0.001). Thus, inflammation alone could not explain the high prevalence of elevated TfR in the present study. This high prevalence might be the consequence of using an inappropriate cut-off: a cut-off of 8.3 mg/L might actually be too low, as suggested for African populations, for whom a higher cut-off of 9.4 mg/L has been proposed 210 . Yet, even with this higher cut-off, 33% of the school children in the present study had an elevated TfR at baseline. Vitamin A fortified rice was very effective in improving vitamin A status. After 6 months of intervention, whereas the prevalence of low vitamin A status increased in the placebo and URO groups who received VA only through the fortified oil included in all types of school meals, it declined in both groups receiving school meals with rice fortified with VA (URN and Nutririce groups). Consumption of Nutririce and URN reduced by 76% and 80% respectively the risk of having marginal VA status when compared with the placebo group receiving unfortified rice. Vitamin A status was also an important predictor of anemia, with the prevalence of anemia almost twice higher for children with marginal VA status or VAD (corrected RBP<1.05 μmol/L) than children with normal VA levels (24.5% vs. 15.0%). Interestingly, there appears to be an effect of the vitamin A containing rice on iron status also, as the increase in FER is found only in the URN and Nutririce groups. Indeed, although URO rice contained the highest concentration of iron, there was no increase in iron stores, whereas the highest increase in FER was found in the rice with the highest vitamin A content (URN). Vitamin A has been shown to increase iron mobilization from stores 372-374 and to improve erythropoiesis. However, it appears that in the present study, erythropoiesis was increased without mobilization of additional iron from stores, given the higher TfR concentrations in the 2 VA containing fortified rice groups. In addition, the VA could have enhanced iron absorption from the gut 375,376 Hence, this study showed that a multi-micronutrient fortified rice containing VA was very effective in improving VA status of school children. However, the effectiveness in improving

109 hemoglobin concentrations and iron status was limited, partly by sub-clinical inflammation. Most of the children had repleted iron stores, yet half of them had elevated TfR. This suggested functional ID and impairment in mobilization or transport of iron from stores to the cells, possibly due to inflammation or other concurrent micronutrient deficiencies like vitamin A, B12 or folate. This study also demonstrates that tackling anemia and micronutrient deficiencies might be optimized by combining fortification strategy with non-nutritional approaches which address infections and inflammation. The impact of this intervention study on anthropometry, cognitive outcomes and zinc and iodine status will be addressed in separate publications.

110

4.2. Effect of fortified rice on cognitive performance in Cambodian school-aged children depends on premix composition and cognitive function tested (Manuscript, drafted)

To be submitted

Marion Fiorentino a, Marlène Perignon a, Khuov Kuong e, Richard de Groot o, Megan Parker g, Kurt Burja f, Marjoleine A Dijkhuizen n, Sek Sokhom h, Chhoun Chamnan e, Jacques Berger a, Frank T Wieringa a

Introduction

In South-East Asia, micronutrient deficiencies remain highly prevalent 3. In addition to the most vulnerable groups (i.e. pregnant and lactating women, and young children), the prevalence of micronutrient deficiencies is also high among school-aged children. Over 30% of school-aged children in South-East Asia are affected by zinc deficiency, while 20% of school-aged children are iron or vitamin A deficient 17 . During the primary school years, anemia and deficiencies of iron, zinc, iodine, vitamin A, vitamin B12, vitamin B6 or vitamin B9 can impair concentration and cognitive function and reduce school attendance by increasing morbidity 17,51,377,378 . Therefore, micronutrient deficiencies are detrimental to optimal schooling, while at the same time education is recognized as a prime opportunity to break the cycle of poverty and undernutrition 379,380 . Although the ‘1000 -days-window’ concept rightly highlights that micronutrient deficiencies in fetal life or early childhood can permanently damage cognitive function and development 381 , it is important to realize that it is still possible to improve cognition among primary school-aged children by improving micronutrient status. Several studies have clearly shown a beneficial impact of micronutrient supplementation or food fortification on morbidity and cognitive performance 53,68,378,382,383 with multiple micronutrients appearing to be more efficient than single or double micronutrient interventions 384 . However, the overall evidence remains equivocal 298,357 . In South East Asia, a region where refined rice typically provides 70% of the dietary energy intake, rice fortification could be a cost-effective strategy to reduce micronutrient deficiencies and anemia 130,175,176,359,360,385 and therefore possibly improve cognitive performance in school-aged children. The school meal program (SMP) of the UN World Food Program (WFP) reached almost 26 million children worldwide in 2011 29 . The primary objective of the WFP SMP is to increase school enrollment and attendance, however, it can also serve to improve child nutritional status at the same time 29 . Fortification of the school meal with micronutrients could help reduce the prevalence of micronutrient deficiency at very little additional cost. For example, depending on technologies, to fortify rice in school meals would only add 0.50 -2.00 US dollar per child/year 178 . In Cambodia, malnutrition among children under five years of age is highly prevalent, with 40% stunting and 11% wasting prevalence 386 . Unfortunately, data on the nutritional status of primary school-aged children are scarce in Cambodia. The WFP distributes school meals to more than 500,000 school-aged children in rural Cambodia 29 . The standard school meal, eaten as breakfast, consists of rice cooked with chick peas, served with canned tomato sauce with fish. Rice is the main component of the program and is also the main vehicle in many other food-

111 based social safety net programs in Cambodia as well as in Asia. Therefore, evidence for the effectiveness of rice fortification would be of immediate importance to these programs as well as for the development of national fortification guidelines in the whole region. The FORISCA project (Fortified Rice for School children in Cambodia) aimed to assess the effect of fortifying the rice in the WFP-SMP with multiple micronutrients on micronutrient status and functional outcomes. The present paper reports on the effects of consuming rice fortified with multiple micronutrients using 3 different fortification formulations, on cognitive performance based on specific cognitive functions in primary school-aged Cambodian children.

Subjects and methods

Study site

The study was conducted in Cambodia, in 16 primary schools from Kampong Speu province, around 60 km south-west of the capital Phnom Penh. Kampong Speu province can be characterized as rural, with most families involved in rice farming. In 2010, prevalence of stunting and underweight in pre-school children in Kampong Speu province was 44% and 9% respectively 386 and in 2005 more than 60% of pre-school children were anemic in Kampong Speu province 387 . Representative data on school children were not available.

Study design, selection of participants and ethics The study period was November 2012 to June 2013, with a rolling recruitment of 1 month and a 6 month intervention period. The study was a double-blind, cluster-randomized, placebo- controlled feeding trial, conducted in 16 schools selected from all primary schools participating in the SMP of WFP (N=18) in Kampong Speu province that gave daily breakfast at school, but did not give Take Home Rations (THR). The sixteen schools were randomly allocated to 1) placebo (normal rice) 2) UltraRice®Original 3) UltraRice®New and 4) NutriRice®. To ensure blinding, the intervention groups were coded with letters (A – H), and 2 schools were allocated to each intervention letter code. Randomization of schools was done by a researcher not involved in the study implementation (MAD), and codes were only known to the researcher and the head of logistics department of WFP, responsible for the distribution of the rice to the schools. Codes were broken only after data collection was completed, and all biochemical data analysis had been done. Before the study commenced, all parents of children of the selected schools were invited to attend a meeting at which objectives and proceedings of the study were explained, as well as their right to refuse to participate in or to continue the study. Only children for whom written informed parental consent was obtained were eligible for randomization. In each school, 132 children were randomly selected from lists provided by the school director, with stratification by grade and gender, giving a total number of 2112 eligible children. Age of each child was calculated using the birthdates provided by the registration book at school, itself based on birth certificates. Exclusion criteria were age < 6 years and severe anemia (hemoglobin concentration <70 g/L 198 ). Severely anemic children received appropriate treatment. The study was approved by the National Ethic Committee for Health Research (NECHR) of the Ministry of Health (Phnom Penh, Cambodia), and the Research Ethics Committee (REC) of PATH (Program for Appropriate Technology in Health, Seattle, USA). The trial was registered at linicalTrials.gov (Identifier: NCT01706419).

112

FIGURE 7 STUDY DESIGN Intervention

The meal was composed of rice (115 g uncooked) and yellow split peas (15 g uncooked) cooked together with a sauce made from canned fish (15 g) , vegetable oil fortified with vitamin A and vitamin D (5 g) and iodized salt (3 g). Fortified rice kernels were produced from fortified rice flour by extrusion, using cold extrusion techniques for UltraRice®Original, warm extrusion techniques for UltraRice®New, and hot extrusion techniques for NutriRice® (DSM). Kernels were blended at a ratio of 1:100 with local normal rice to produce fortified rice. The same rice was used for the placebo intervention. The micronutrient composition of the UltraRice®Original (URO), UltraRice®New (URN), and NutriRice® (NR) is provided in Table 28. The meal was cooked in the school kitchen and served as breakfast at the beginning of the school day. The school meal was distributed for at least 6 months during school days (6 days

113 per week, except during national holidays). An earlier study had shown that fortified rice (UltraRice®Original and NutriRice®) were highly accepted by Cambodian school-aged children 179 . Children were dewormed using mebendazole after baseline and endline sample collection. Cognitive performance, anthropometry, parasite infestation and micronutrient status were evaluated at baseline and after 6 months of intervention

TABLE 28 MICRONUTRIENT COMPOSITION OF UNCOOKED RICE PER 100 G OF BLENDED RICE

Vitamin Vitamin Vitamin Vitamin Vitamin Iron Zinc Vitamin B6 B12 B1 B9 B3 (mg) (mg) A (IU) (mg) (ug) (mg) (mg) (mg) UltraRice®original 10.7 3.0 0.0 0.0 0.0 1.1 0.2 0.0 Ultr aRice®new 7.6 2.0 2140.0 0.0 3.8 1.4 0.3 12.6 NutriRice® 7.5 3.7 960.0 0.9 1.3 0.7 0.1 8.0

Data collection

Blood samples and anthropometric measurements were collected and cognitive tests performed at schools, after breakfast between 7:30 am and 10:00 am.

Cognitive tests Cognitive tools generally suffer from intercultural variability 298,388 : in order to avoid bias linked to translation in Khmer language, tests not involving oral items and with predominantly pictorial content have been chosen in the present study 389 . Three tests were performed to assess cognitive performance: the Raven’s Colored Progressive Matrices (RCPM), a common non-verbal test of overall intellectual ability in children ≥ 5 y 37,306,390 , and two standardized tests from the Wechsler Intelligence Scale for Children (WISC III) widely used to evaluate the intelligence of children aged 6-16 y 307 : block design and picture completion.. RCPM and WISC III tests were previously used in Vietnamese school children in nutritional intervention trials 391 .RCPM consists of selecting the correct piece (choice of 6 different patterns) that completes an illustrated pattern. The Block design test requires constructing a design with blocks to match a given design on a picture. The Picture completion test consists of identifying the missing detail in a picture. Higher scores indicate better performance. Fluid intelligence underlies reasoning functions whereas crystallized intelligence is related to experience, education and culture as it requires acquired skills and knowledge 383 . Block design and RCPM tests are associated with fluid intelligence 392-394 while picture completion is related to crystallized intelligence 395 . Each child was tested individually. The cognitive tests were administered by a team of 25 students from the Psychology Department of Royal University of Phnom Penh. The team was specifically trained during a one week workshop prior to the start of the intervention to ensure standardization of proceedings and scoring.

Anthropometric measurements Weight and height were measured without footwear and wearing minimal clothing, using standardized procedures 159 . Weight was measured once to the nearest 100 g (Seca 881 U scale, Germany). The accuracy of the scales was checked every day using a set of 2 calibration weights. Height was measured twice to the nearest 0.1 cm on a wooden stadiometer and mean values were used. When differences between two measures of height for the same child exceeded 0.5 cm, measurements were repeated. Height-for-age (HAZ) and BMI-for-age z-scores (BAZ) were

114 calculated according to the WHO 2006 reference 157 . Stunting and severe stunting were defined as HAZ<-2 and HAZ<-3, respectively. Thinness and severe thinness were defined as BAZ<-2 and BAZ<-3, respectively. Overweight was defined by 12 z -scores. Anthropometric measurements were performed by 3 teams who were trained and standardized before baseline and again before endline 159 .

Blood and urine sample collection and hemoglobin concentration measurements Five mL of venous blood were collected by experienced phlebotomists in a Vacuette (Greiner Bio One) trace-element free vacutainer without anticoagulant. Urine was taken in a sterile container. Urine and blood samples were stored immediately at 4 °C in an icebox containing ice- packs and transported to the laboratory within a maximum of 5 hours after the first sample was obtained. Blood samples were centrifuged at 2700 G for 10 minutes at room temperature. Serum and urine where then aliquoted in Eppendorf tubes and stored at -30°C. Hemoglobin concentration (Hb) was measured on site in whole blood immediately after blood taking, using HemoCue® R 301+ and HemoCue controls (Hemotrol low, medium, high, HemoCue ®).

Parasite infestation Plastic containers and instructions for stool sample collection were given to the child on the day of data collection and requested to be returned to the school the following day. Stool samples were then stored in a cool box, transferred to the National Malaria Center (CNM, Phnom Penh, Cambodia) and stored at 4°C until analysis. Quantitative parasite eggs counts were performed by CNM using the Kato-Katz method 314 .

Laboratory analysis

Ferritin (FER), soluble transferrin Receptor (TFR), retinol-binding protein (RBP), C-reactive protein (CRP), α1 -acid-glycoprotein (AGP) serum concentrations Serum samples were sent on dry ice to the VitMin laboratory (Willstaett, Germany) for determination of RBP, CRP, FER, TfR, and AGP concentrations. All these proteins were measured by a sandwich enzyme-linked immunosorbent assay (ELISA) technique 195 . Inflammation was defined as an elevated CRP (>5 mg/L) and/or elevated AGP (>1 g/L) allowing differentiation between incubation phase (high CRP), convalescence phase (both AGP and CRP elevated) and late convalescence phase (elevated AGP only) 197 . Anemia was defined by Hb below cut-offs depending on age and gender: 115 g/L for participants <12 y; 120 g/L for adolescents between 12 and 15 y and girls ≥15 y; 130 g/L for boys ≥15 y; severe anemia wa s defined as Hb<70 g/L; depleted iron stores were defined by inflammation-corrected FER<15 g/L 198 . Correction factors of FER were 0.77, 0.53 and 0.75 for participants respectively in incubation, early convalescence, and late convalescence phases, after Thurnham et al 197 . Tissue iron deficiency was defined by TfR>8.3 mg/L 195 . Low FER and high TfR concentrations are both considered indicators of iron deficiency (ID) 198 so ID was defined by depleted iron stores (low FER) and/or iron tissue deficiency (high TfR). Body iron was calculated according to the formula of Cook 199 : body iron (mg/kg) =2(log (TfR/FER ratio)22.8229)/0.1207, using FER corrected for inflammation. Body iron was considered low when < 4 mg/kg 396 . Vitamin A status was measured using RBP concentration which reflects serum retinol concentration as RBP occurs in a 1:1:1 complex with retinol and transthyretin 200 . RBP concentrations were corrected in participants with inflammation by factors 1.15, 1.32, 1.12 respectively for incubation, early convalescence and late convalescence 201 . Vitamin A deficiency (VAD) was defined by corrected

115

RBP<0.7 mol/L 200 and marginal VAD by cor rected RBP values ≥0.7 mol/L and <1.05 mol/L 202 .

Serum zinc concentrations Serum samples were sent on dry ice to the National Institute of Nutrition (Hanoi, Vietnam). Serum zinc was measured by flame atomic absorption spectrophotometry (AAS), using trace- element free procedures. Considering that none of the children were fasting due to the school breakfast, zinc deficiency was defined by serum zinc concentration <0.65 mg/L for participants <1 0 y ;for participants ≥10 y, cut -offs are 0.66 mg/L for girls and 0.70 mg/L for boys 203 .Severe zinc deficiency was defined as serum zinc concentration <0.5 mg/L 397 .

Urinary iodine concentration Urine samples were sent on dry ice to the National Institute of Nutrition (Hanoi, Vietnam). Urinary iodine was measured using an ammonium persuflate method at the Thai Nguyen Provincial Hospital, Vietnam, which is one of the 3 reference laboratories for iodine determination in Vietnam 196 . Iodine deficiency and iodine status above requirements were respectively defined by urinary iodine concentration (UIC) below 50 µg/L and above 200 µg/L 124 .

Rice sample analysis Composition of the fortified kernels for iron, zinc, vitamin A, B6, B12, B9 and B3 was analyzed after production by Silliker laboratories using standard methods. The final composition of blended fortified rice was calculated taking into account a mixing ratio of 1:100. The composition of the fortified rice kernels is given in Table 28.

Data management and statistical analysis Data entry, including quality checks and validation by double entry of questionnaires, was performed with EpiData version 3.1 (EpiData Association, Odense, Denmark). Data management and analyses were performed using SPSS version 20.0 software (SPSS, Inc., Chicago, USA). Normality of data was checked before analysis with the Kolmogorov-Smirnoff test. In addition, absolute values of skewness and kurtosis of the distribution curves of<2 and <7 were used to indicate that the distribution of continuous variables was close to normal 398 . Baseline characteristics were compared between intervention groups using ANOVA for continuous variables, and using Pearson χ2 test for categorical variables.

Primary analysis: Generalized linear mixed models including interventions groups, visit (baseline, endline), intervention groups x visit, age, gender and clustering (school within group) as fixed factors were used to assess the impact of intervention on the raw scores of the cognitive tests compared to the placebo group. P-value <0.05 was considered as significant and P-value between 0.1 and 0.05 as a tendency.

Secondary analysis : Low iron status, helminthes infection and stunting were identified earlier as risk factors for low cognitive scores in children participating in the FORISCA study 399 . Inflammation was shown to significantly affect the effect of the FORISCA intervention on micronutrient status 400 In order to assess if these covariates had an impact on the effect of the interventions on cognitive scores in this study, interactions of body iron below 4 mg/kg, helminthes infection, stunting and inflammation were included as fixed factors in the previous generalized mixed models.

116

Results

Baseline characteristics

In total cognition data were available for 1933 children at baseline and 1796 children at endline (Figure 7). Biochemical, physiologic and anthropometric characteristics of the 1933 children at baseline are presented in Table 29. Mean age at enrollment was 9.7 years old, with participants’ age ranging from 6 years to 16 years. Due to late school enrollment and re peating of class, 44% of the participants were ≥ 10 years, defined as adolescent according to the WHO definition 401 , but in this paper all participants will be referred to children. Stunting was highly prevalent in the children participating in the FORISCA study (43 %) whereas 26% classified as ‘thin’. Iron stores were adequate (FER>15 g/L) in most children (99%), whereas paradoxically, tissue iron deficiency was highly prevalent (48% of participants with TfR>8.3 mg/L). Accordingly, half of the children were classified as iron deficient, but this was almost exclusively related to high TfR. Almost all of the children were zinc deficient (>90%) while less than 1% of the children were vitamin A deficient. Prevalence of anemia was 17%.

Primary analysis

All cognitive scores improved over the 6 months intervention (P<0.001, Table 30). On average, the scores increased by 7 (54%), 3 (20%) and 3 (38%) points for block design, RCPM, and picture completion tests respectively. The intervention had a significant overall impact on block design scores (P=0.003). Improvement of block design scores were significantly higher comparatively to t he placebo group only in children consuming UltraRice®original (β=1.17, P=0.03). No significant difference in RCPM scores or picture completion scores was found between the intervention groups.

Secondary analysis

For block design scores, only stunting had an influence on the impact of the intervention (P=0.006 for overall interaction, Table 31) while intestinal parasite infection, inflammation and low body iron did not. The increase in block design score over time was higher in non-stunted children receiving NutriRice® compared to stunted children receiving NutriRice® (difference of scores -0.69, P=0.05). For RCPM stores, parasite infection had a negative effect on the intervention (P=0.045 for overall interaction). Among children receiving UltraRice®Original, RCPM scores increased more in children without parasites than in children with parasites (difference of scores -0.08, P=0.010). Low body iron, stunting and inflammation did not affect the impact of the intervention with regard to RCPM scores. For picture completion scores, both stunting and low body iron had a strong effect on the impact of the intervention (P<0.001 and P=0.001 respectively). In children receiving UltraRice®New, those with low body iron increased less in their picture completion scores compared to children with body iron ≥ 4 mg/kg (difference of scores 0.51, P=0.001). Similarly the scores of stunted children increased less compared to non-stunted children receiving UltraRice®Original (difference of scores -0.51, P=0.015).

117

TABLE 29 CHARACTERISTICS OF CHILDREN AT BASELINE (WITH AVAILABLE COGNITION DATA )

p- n ALL Placebo URO URN NR value % girls 50.3 50.7 50.1 50.7 49.8 NS 1933 age (years) 9.7 ± 2.2 9.6 ± 2.3 9.6 ± 2.2 9.7 ± 2.2 9.8 ± 2.4 NS % stunted children HAZ < - 2 z-score 42.6 42.9 39.3 45.0 45.1 NS 1927 % thin children BAZ < - 2 z-score 25.8 24.9 26.7 29.8 21.9 0.040

% children infected by helminthes 20.3 23.0 24.9 19.8 12.9 0.000 1442 % children lightly infected by hookworms 17.8 20.0 22.2 17.4 11.1 0.001

% children with inflammation (CRP> 5 mg/L 1882 or AGP > 1g/L) 37.8 43.2 42.8 32.8 32.7 0.000 hemoglobin (g/dL) 1903 12.4 ± 1 12.4 ± 1 12.4 ± 1 12.4 ± 0.9 12.4 ± 1.1 NS % anemic children * 1903 17.1 18.4 15.4 16.9 17.6 NS 79.8 ± 70.1 ± 71.9 ± ferritin (mg/L) ** 74.8 ± 37.3 77.5 ± 38 36.6 35.2 37.3 0.000 % children with low ferritin (ferritin corrected for inflammation <15 μg/L) 1.7 0.4 0.6 3.7 1.9 0.001 TfR (mg/L) 8.9 ± 2.7 9 ± 2.6 9.1 ± 2.6 8.5 ± 2.5 8.9 ± 3.1 0.001 % children with high TfR (TfR > 8.3 mg/L) 1882 47.8 43.7 43.5 52.4 51.4 0.003 % children with iron deficiency anemia (anemia with high TfR and/or low ferritin) 7.7 6.5 5.6 9.5 9.0 0.096 total body iron (mg/kg) 5.9±2.3 6±2 6±2.2 5.8±2.4 5.8±2.5 0.061 % children with low iron status (total body iron < 4 mg/kg) 15.1 12.9 13.5 15.9 18.1 NS zinc (umol/L) 7.6 ± 1.7 7.2 ± 1.6 7.8 ± 1.7 7.5 ± 1.7 7.8 ± 1.7 0.000 1546 % zinc deficient children *** 93.2 98.1 91.2 93.7 91.9 0.000 % severely zinc deficient children **** 53.7 51.1 48.9 55.5 51.1 0.000 iodine (ug/L) 32.8 22.4 25.3 49.0 33.6 0.000 % children < 50 ug/L 1824 17.2 6.4 9.0 35.0 17.4 0.000 % children >200 ug/L 0.3 0.0 0.0 1.3 0.0 0.001

RBP (mmol/L) 1.6 ± 0.4 1.6 ± 0.4 1.7 ± 0.4 1.5 ± 0.4 1.5 ± 0.4 0.000 % vitamin A deficient children (RBP< 0.7 umol/L) 1882 0.9 0.6 0.2 1.2 1.5 NS % children with marginal vitamin A status (RBP<1.05 umol/L) 9.6 7.3 3.6 14.2 12.9 0.000 * Hb < 115 g/L for participants <12 y; 120 g/L for teenagers between 12 and 15 y and girls ≥15 y; 130 g/L for boys ≥15 y ** FER corrected for inflammation <15 mg/L *** serum zinc <0.65 mg/L for participants <10 y ; for participants ≥10 y, cut -offs are 0.66 mg/L for girls and 0.70 mg/L for boys **** serum zinc <0.5 mg/L NS : Non Significant

118

TABLE 30 COGNITION OUTCOMES AND EFFECT SIZES AFTER 6 MONTHS OF INTERVENTION FOR ALL CHILDREN

Block design score Effect of variables Interaction term*

P value mean SE β coefficient [95% CI] P-value

group 0.075 Baseline

age 0.000 UltraRice®original 13.6 0.4 - - gender 0.000 UltraRice®new 14.1 0.4 - - visit 0.000 NutriRice® 13.6 0.4 - - school(group) 0.000 placebo 13.1 0.4 - - groupxvisit 0.003 Endine UltraRice®original 22.0 0.5 1.17 [ 0.12 ; 2.22 ] 0.029 UltraRice®new 20.5 0.5 -0.66 [ -1.71 ; 0.38 ] 0.214 NutriRice® 20.4 0.5 -0.45 [ -1.50 ; 0.60 ] 0.396 BIC** 25971 placebo 20.5 0.5 - -

RCPM score Effect of variables Interaction term*

P value mean SE β coefficient [95% CI] P-value

group 0.000 Baseline

age 0.000 UltraRice®original 17.5 0.2 - - gender 0.000 UltraRice®new 17.1 0.2 - - visit 0.000 NutriRice® 16.4 0.2 - - school(group) 0.000 placebo 16.6 0.2 - - groupxvisit 0.153 Endine UltraRice®original 21.0 0.2 0.02 [ -0.52 ; 0.57 ] 0.930 UltraRice®new 20.1 0.2 -0.34 [ -0.87 ; 0.20 ] 0.221 NutriRice® 19.9 0.2 0.29 [ -0.25 ; 0.82 ] 0.300 BIC** 20736 placebo 20.1 0.2 - -

Picture completion score Effect of variables Interaction term*

P value mean SE β coefficient [95% CI] P-value

group 0.076 Baseline

age 0.000 UltraRice®original 7.9 1.9 - - gender 0.078 UltraRice®new 7.5 1.8 - - visit 0.000 NutriRice® 7.8 1.8 - - school(group) 0.000 placebo 7.6 1.9 - - groupxvisit 0.094 Endine UltraRice®original 11.0 1.9 0.09 [ -0.47 ; 0.64 ] 0.760 UltraRice®new 10.9 1.9 0.34 [ -0.21 ; 0.89 ] 0.230 NutriRice® 10.4 1.9 -0.36 [ -0.91 ; 0.19 ] 0.204 BIC** 20640 placebo 10.7 1.9 - -

Results from generalized mixed models including interventions groups, visit (baseline, endline), intervention groups x visit, age, gender and clustering (school within group) as fixed factors

* interaction term groupxvisit(=endline) ** Bayesian Information Criterion

119

TABLE 31 EFFECTS OF RISK FACTORS FOR LOW COGNITIVE SCORES ON THE 6 MONTHS INTERVENTION

Overall effect of interaction UltraRice®original UltraRice®new NutriRice® with the intervention *

difference difference difference P- P- P- β in scores β in scores β in scores P-value value value value coeff* increase coeff* increase coeff* increase *** *** *** ** ** **

Stunting Block design 0.006 -1.411 0.42 0.234 -1.805 0.93 0.144 -2.687 -0.69 0.05 Picture completion 0.000 -1.146 -0.51 0.015 -0.434 0.68 0.374 -0.243 0.91 0.654 RCPM 0.352 -0.774 0.14 0.147 -1.131 -0.29 0.042 -0.566 -0.21 0.358

Parasite Block design 0.197 3.152 -1.18 0.009 -0.352 -0.18 0.785 0.935 -0.31 0.524 infection Picture completion 0.129 -0.389 -0.29 0.411 0.407 -0.82 0.428 -0.072 -0.32 0.902 RCPM 0.045 -1.389 -0.08 0.01 -0.121 -0.66 0.835 -0.532 0.03 0.421

Inflammation Block design 0.141 -1.130 -0.80 0.314 0.471 -0.67 0.706 -0.136 2.08 0.919 Picture completion 0.873 0.275 0.35 0.534 -0.436 0.00 0.379 -0.01 0.18 0.985 RCPM 0.058 0.388 0.38 0.442 -0.283 0.04 0.615 0.09 1.30 0.881

Low body iron Block design 0.157 1.079 0.06 0.453 2.328 -2.91 0.103 -1.582 0.67 0.308 Picture completion 0.001 -1.011 0.50 0.076 -1.853 -0.51 0.001 0.607 -0.63 0.323 RCPM 0.211 -1.144 -0.44 0.077 -0.971 -0.66 0.13 -0.102 -0.29 0.884 Results from generalized mixed models including age, gender, clustering (school within group), interventions groups, visit (baseline, endline), intervention groups x visit, and each of risk factor (stunting, parasite infection, inflammation or low body iron) x intervention groups x visit, as fixed factors * β coefficient of interaction term group x visit x risk factor (stunting, parasite infection, inflammation or low body iron) ** Difference of improvement of scores between baseline and endline in stunted children compared to non-stunted children (idem for children with parasites compared to children without parasites, children with inflammation compared to children without inflammation, children with low body iron compared to children with normal body iron) *** Significance of the difference of improvement of scores 120

Discussion

To our knowledge, this is the first paper reporting modest but significant effects of a micronutrient fortified rice provided in a school meal program on cognitive function in primary school children. Beneficial effects of UltraRice®Original on micronutrient status and anemia were reported in small populations (n=210) of Mexican women 360 , Indian school-aged children 175 and Thai school-aged children 177 . No effect of NutriRice® was found on Indian children’s cognitive function, but the sample size was considerably smaller than in the present study 130 . Duration of these studies was comparable to the FORISCA study, with meals containing UltraRice®Original or NutriRice® being served 5-6 days for 5 to 8 months. Perhaps surprisingly, the present study found only an effect of UltraRice®Original on cognitive performance, and not of the two other types of fortified rice. UltraRice®Original contained the highest concentration of iron, which is known to be important in brain development, also during pre-adolescence and adolescence. Iron plays a role in neurotransmission, especially in the dopamine pathway, which serves in memory, learning, attention, motor control, and emotional affect modulation 36,60 . Increasing iron intake in iron deficient and/or anemic children using fortification or supplementation has been shown to improve schooling or cognitive outcomes like learning, memory, concentration or school achievement 53,54,382,402-405 . Zinc is also involved in neurotransmission pathways 406 . Moreover, zinc as a component of zinc-dependent enzymes, plays an important role in neuronal genesis and migration, making adolescents particularly sensitive to zinc deficiency due to rapid brain growth, similar to early childhood 378,406,407 . Even though data are lacking about long-term effects of zinc treatment on cognitive performance, there is some evidence that zinc supplementation may improve neuropsychological functions 378 especially reasoning 408 . For example, zinc supplementation had beneficial effects on cognitive scores of Indian adolescent girls, by reducing reaction time, improving memory and RCPM scores 82 . However, a recent meta-analysis found no effect of zinc on cognitive functioning, but reported that there is a lack of high quality trials 409 As reported elsewhere, the intervention had a modest but significant and consistent impact on the micronutrient status of children participating in the FORISCA study 280,400 : A significant increase of FER, indicating improvement of iron stores, between baseline and endline was observed only in children receiving UltraRice®New and NutriRice® (P<0.001). However, at the same time TFR significantly increased in children receiving UltraRice®New and NutriRice® (P<0.001), indicating decreasing iron tissue stores. In contrast, TFR tended to decrease in the UltraRice®Original group (P=0.088), indicating improved iron tissue stores. Hence, although it did not improve iron stores UltraRice®Original was the only fortified rice which improved iron tissue status. It is tempting to link the improved iron tissue status in the UltraRice®Original to the improvements of cognitive performance found in this group. And the results also suggest that TFR would be a better indicator for functional iron deficiency than FER, raising questions on which indicator should be used for assessing iron status. In an overview of randomized controlled trials of multiple micronutrient supplementation in school-aged children, Eilander et al. noticed a potential beneficial effect on fluid intelligence and school performance, but not on crystallized intelligence 383 which is consistent with fortified rice having an impact on block design scores but not on picture completion scores in children participating in the FORISCA study. The block design test measures perceptual reasoning and executive functions, both of which are associated with frontal lobe functions 395,410 . The development of the full range of executive functions is suggested to occur in late childhood and adolescence as myelination of the frontal lobes proceeds 37 with peaks at 7, 9 and 12 years of

121 age 36,38 .There is some evidence that iron is a key nutrient in the development of these executive functions 36 . Although UltraRice®Original contained the highest iron content, the difference with the 2 other types of fortified rice was relatively small, and cannot completely explain the observed difference in impact on cognition. Therefore, it is likely that interactions with other micronutrients present in the UltraRice®New and NutriRice®, but not in the UltraRice®Original, also play a role in modifying the effect on cognitive outcomes. UltraRice®New had the lowest concentrations of zinc and iron but the highest of vitamins (A, B6, B12, B1, B9, and B3). Even though vitamin B12, B6,and B9 are known to play a role in cognitive performance by being involved in methylation in the central nervous system 411,412 , UltraRice®New and NutriRice®, fortified with these vitamins, did not improve cognition performance in the present study while, UltraRice®Original, not fortified with these vitamins, did improve cognition performance. Therefore, improvement in cognition performance in the present study is more likely to be related to the additional iron and zinc provided through UltraRice®Original, and not to additional B-vitamins.

Although the impact of UltraRice®Original on RCPM scores did not reach significance, it was significantly lower in children with intestinal parasite infection than in those without parasites. Hence, while being a risk factor in itself for poor cognitive scores, parasite infestation also limited the impact of fortified rice on RCPM scores. Earlier we reported that fortified rice also increased parasite infestation in children participating in the FORISCA study 413 , highlighting the complex interaction between nutrition and infection.

Albeit no significant impact of fortified rice on RCPM and picture completion scores was observed on the general sample, low body iron diminished the increase of these scores over the intervention period regardless of intervention type, indicating that children participating in the FORISCA study with low iron status benefited less from fortified rice with respect to cognition scores, in comparison to children with adequate iron status. Even if increased iron intake during the intervention had an overall beneficial effect on block design scores, this suggests that the amount of bio-available iron in the fortified rice may not be sufficient to reverse the negative effect of iron deficiency on cognition scores.

Inflammation also had a negative effect on RCPM scores improvement. Interestingly, this is in parallel with the observed effects of fortified rice on iron status of children participating in the FORISCA study, which emerged stronger after removing children with inflammation from the analysis 400 . This suggests that the acute-phase-response may disturb the response to fortified foods and therefore the impact on functional outcomes such as of cognitive performance.

Strengths and limitations of the study

One of the strengths of this study was its design as an effectiveness trial. We assessed the impact of the introduction of fortified rice within an existing program, with minimal interference in the program’s daily logistics. Hence, the results obtained pr ovide a realistic indication of what can be expected from the introduction of multi-micronutrient fortified rice on cognitive performance in Cambodian school-aged children. Another strength of the study is the sample size, with >2000 children being followed over the study period. Also, cognitive tools generally suffer from intercultural variability 298,388 but, in order to avoid bias linked to translation in

122

Khmer language, tests not involving oral items and with predominantly pictorial content have been chosen in the present study 389 .

One limitation of this s tudy was the intervention’s short duration of only 6 months; this amount of time may not have been enough to improve cognitive performance to its full potential. An intervention exceeding 6 months may have shown a larger impact on cognitive outcomes. Also, the present study was an effectiveness study, not an efficacy study. Hence, quantities of rice consumed every day were not controlled, nor was the micronutrient status of children enrolled in the study. The prevalence of low iron stores was very low, and larger impact on cognitive function may have been observed in a population with a higher prevalence. Although the schools were randomly allocated to the different intervention groups, groups still differed with respect to thinness, iron status, iodine status, micronutrient deficiencies and parasite infestation. Groups did not differ with respect to gender, age, and stunting. As the subject size is large, and the statistical modeling robust, no attempts were made to correct or amend for this, instead the validity of each analysis was carefully checked. In addition, the pre-post comparison model of the study design further aids the clear interpretation of the data. Another limitation of the FORISCA study is that levels of iron and zinc used in the present study are below the current recommended levels for fortified rice, which might have reduced the effectiveness 180 .

Conclusion The SMP is known to be an incentive for school attendance and enrollment, and could enhance learning by reducing hunger and increasing concentration 29 .The current study illustrates that fortification of school meals with multiple micronutrients can improve cognitive performance, and thus promote schooling further. However, the improvements in cognitive performance were modest and impact was modified by several external factors such as stunting, intestinal parasite infestation, low body iron and inflammation. This strengthens the case for combining health interventions, such as deworming combined with provision of micronutrients. However, micronutrients used in rice fortification should be carefully selected and appropriately dosed, taking for example the initial nutritional and infection status of the target population also into account.

123

124

Chapter 5. Evaluation of currently used indicators for malnutrition using data from Cambodia and Senegal

125

5.1. Current MUAC cut-offs to screen for acute malnutrition need to be adapted to gender and age: the example of Cambodia (Published paper)

Published in Plos One (2016)

Marion Fiorentino a, Prak Sophonneary p, Arnaud Laillou q, Sophie Whitney g, Richard de Groot o, Marlène Perignon a, Khov Kuong e , Jacques Berger a , Frank T Wieringa a

Introduction

Wasting or acute malnutrition is a major contributor to the global disease burden and to child mortality. In 2011, >50 million or 8% of all children <5y (years) were affected 3,18 . Acute malnutrition can be divided into severe acute malnutrition, being defined as a weight-for-height Z-score (WHZ) <-3 or a mid-upper arm circumference of <11.5 cm, or moderate acute malnutrition, with a WHZ between -2 and -3 Z-scores or a MUAC (mid-upper-arm circumference) between 11.5 and 12.5 cm. Severe acute malnutrition affected 19 million children in 2011, causing an estimated 500,000 deaths, which represents ~7.5% of all<5y mortality 3. For early identification of children with acute malnutrition, an easy, accurate, and low-cost indicator is needed. The golden standard to identify acute malnutrition is weight-for-height z- scores 414 . Unfortunately, scales and/or height board are not always available for screening at community level in many developing countries. Therefore, as subcutaneous fat and muscle mass decrease in undernourished children, MUAC has been used as a proxy indicator to screen for acute malnutrition. Moreover, MUAC was shown to predict child mortality at least as well as WHZ 415-417 . In the revised (2013) guidelines for the management of severe acute malnutrition, the World Health Organization (WHO) recommends using 11.5cm and 12.5cm as cut-offs for admission and discharge criteria for severe and moderate acute malnutrition in children under 5y respectively 418 . However, recently we showed that MUAC and WHZ identify different groups of children with malnutrition, and that a cut-off of 13.3 cm, rather than 11.5 cm would increase sensitivity of MUAC to identify children with a WHZ<-2 or <-3 419 . However, it is questionable whether using only 1 cut-off for all children between 6 and 59 months of age is valid, as MUAC was shown to be age and gender dependent in several studies 420,421 . Recent research has highlighted the problems with using one, unique cut-off for MUAC for identifying malnutrition, as it respectively over- and underestimates malnutrition in children under and above 2 years old 421,422 , and identifies more girls than boys with malnutrition. Onis et al (1997) developed a MUAC-for-age z-scores reference based on US data 422 . However, this reference requires an accuracy of the child's age of 1 month. At community level, it is often very difficult to estimate age precisely and a table reference is then needed to verify whether the child is malnourished or not. The latest WHO guidelines for the treatment of severe acute malnutrition recommend refining the cut-offs for MUAC to identify acute malnutrition for different groups in children under 5 y, using perhaps also different cut-offs for stunted and non- stunted children, and establishing cut-offs for children <6 months or >5 y 418 . Malnutrition in the latter age group is important to address also, even though it is less linked to mortality. Acute malnutrition during school years can impair physical and mental development 18 , and acute malnutrition is widely spread also in school-aged children. For example, one third of Southeast Asian school children are affected by thinness 17 .

126

The present study aimed to answer several research questions posed in the revised WHO guidelines for the treatment of SAM, especially focusing on the discrepancy between MUAC and WHZ, defining new MUAC cut-offs adapted to age and gender in children under 5 y, and MUAC cut-offs for older children. We used pooled data from several surveys carried out in Cambodia between 2011 and 2013. Cambodia is one of the 42 countries with the highest child mortality rates 423 , and prevalence of acute malnutrition is high with 10% of <5y children having a weight- for-height Z-score (WHZ) <-2 295 . The main objectives of the study was to find new cut-off values for MUAC in children from 0 to 14 y based on age groups of 0.5 - 2 y, 2 - 5 y, 5 - 8 y, 8 - 11 y and 11 - 14 y, to identify WHZ- scores of <-2 and -3 Z-scores respectively, and to verify whether there were gender differences or differences between stunted and non-stunted children respectively.

Methods

Data on weight, height, MUAC, gender and age were available for a total of 14,157 Cambodian children, from 5 different surveys conducted in Cambodia by several organizations (UNICEF; WHO; WFP; World Vision; International Relief and Development; Institut de Recherche pour le Développement) between 2011 and 2013. All parents of participants gave their written consent. All studies were approved by the National Ethic Committee for Health Research (NECHR) of the Ministry of Health, Phnom Penh, Cambodia. In 4 studies, the target population was children aged 0-4.9 y whereas in 1 study children aged 5-14 years were included.

Measurements of MUAC (to the nearest 1 mm) were made using a non-stretch tape measure (provided through UNICEF) in all studies. Weight was measured with the child wearing only light clothes, and was recorded to the nearest 0.1 kg by a Salter scale in the 4 studies on children under 5y and by a Seca scale in the study on children above 6 y. Length was measured in children under 2 yrs and height was measured in children above 2 yrs. Length/height was measured to the nearest 0.1 cm with a Holtain infantometer in children under 6 y and with a locally produced stadiometer in children above 6 y. Height boards and scales were daily calibrated. Anthropometric teams were standardized 159 . Age was determined (calculated in the nearest months) by asking of both the child’s age and date of birth to mothers of children under 6 y. In the study on children above 6 y, date of birth was collected from school registers. Data were entered into PASW statistics 18 (Chicago, USA) or Epidata Entry (EpiData Association, http://www.epidata.dk). Weight for Height z-scores (WHZ), Height for Age z-scores (HAZ), Body Mass Index for Age z-scores (BAZ) were calculated from anthropometric data using WHO AnthroPlus v1.0.4. As recommended by WHO, and using the WHO child growth references 424 , acute malnutrition was defined in children under 5 y by respectively WHZ<-3 z-score (severe acute malnutrition) and WHZ between -3 and -2 z-scores (moderate acute malnutrition), and in children above 5 y by respectively BAZ<-3 z-scores (severe acute malnutrition) and BAZ between -3 and -2 z-scores (moderate acute malnutrition). Stunting was defined as a HAZ<-2 z- scores.

To assess the performance of MUAC cut-offs compared to the golden standard recommended by WHO to define severe and moderate acute malnutrition), receiver operating characteristic curves (ROC curves) were constructed. The sensitivity and false positive rates (1-specificity) of MUAC were determined using wasting (WHZ<-2 z-score in children under 5 y) and thinness (BAZ<-2 z-score in children above 5 y) as gold standards of acute malnutrition. The ROC curve is the plot of sensitivity versus false positive rate of MUAC cut-offs. The area under curve (AUC) is

127 the area between the curve and the segment (0,0) and (1,1), which corresponds to a random classifier. A larger AUC indicates a more accurate diagnosis of acute malnutrition defined by WHZ cut-offs 425 (17). Data analysis was performed using SPSS version 20.0 (SPSS, Inc., Chicago, IL).

ROC curves were constructed for a large set of MUAC cut-offs, increased by steps of 1 mm, in order to find the new cut-off for severe and moderate acute malnutrition respectively. We generated a dummy variable for each MUAC cut-off, and did this for each MUAC-value in steps of 1 mm. Then, we calculated the AUC for that dummy variable in comparison to the golden standard, using the same methodology as Laillou et al and Fernandez et al 419,426 (8, 18). In order to evaluate the performance of our analysis, the corresponding Youden index, which is the difference between the true positive rate (sensitivity) and the false positive rate, was calculated : 1 indicating a perfect test, and 0 a useless test 427 . These analyses were conducted overall, and by gender and age groups. Several criteria were chosen to select new cut-off presented in this paper. The best correspondence between 2 indicators was first defined as the MUAC cut-off with highest AUC in the ROC curve. However, the sensitivity which one wants to be optimized for screening purposes, in order to miss a minimum of children with acute malnutrition, increased considerably around the MUAC cut-off with the highest AUC, whereas the AUC hardly changed (cf. supporting information figures). Thus, among the highest values of AUC (within 0.02 from highest AUC), the cut-off with the highest sensitivity was selected. As this can lead to many false-positive children to be included after screening, a third criteria was introduced by the authors to define the new cut-off: a false positive rate below 1/3 of non-malnourished children. To summarize, among the highest values of the area under the curve (highest AUC - 0.02), the cut-off with the highest sensitivity and a false positive rate of less than 33% was considered to be the new optimal cut-off. In addition, we calculated the difference of Youden index between the new optimal cut-off and the cut-off obtaining the highest Youden index. These calculations were done for 0.5 - 2 y, 2 - 5 y, 5 - 8 y, 8 - 11 y and 11 - 14 y, for girls, boys and all. Calculations were repeated for stunted and non stunted children under 5 y respectively 418 .

Results

In total, data was available for 14,157 children (51% male), ranging in age from 0 - 14 y. Characteristics of the children are presented in table 32.

Stunting increased with age with respectively 36% and 41% of children under and above 5 yrs aving HAZ <-2 z-scores. The same patterns was seen for acute malnutrition with the prevalence of WHZ<-2 z-scores and WHZ<-3 z-scores being 11% and 1% respectively in children under 5 y, whereas thinness and severe thinness were found in 26% and 5% of the children above 5 respectively. In children under 5 y, prevalence of acute malnutrition and severe acute malnutrition identified using actual WHO MUAC cut-offs of 12.5 cm and 11.5 cm were 3% and 0% respectively (Table 33).

The sensitivity of MUAC ranged from 6.5% to 32.9% in children with acute malnutrition and from 0% to 18.2% in children with severe acute malnutrition with large differences between boys and girls (Table 33).

128

TABLE 32 AGE AND GENDER CHARACTERISTICS OF THE PARTICIPANTS AND PROPORTION OF CHILDREN SUFFERING FROM ACUTE MALNUTRITION

Age Boys Girls All

0-23 months N (%) 2849 (20%) 2736 (20%) 5585 (39%)

WHZ <-2 z-scores (%) 13% 10% 12% 24-59 months N (%) 3147 (22%) 3031 (22%) 6178 (44%) WHZ <-2 z-scores (%) (%)malnutrition* 10% 9% 10% 5-7.9 years N (%) 366 (3%) 379 (3%) 745 (5%) BAZ <-2 z-scores (%) 19% 14% 16% 8-10.9 years N (%) 451 (3%) 519 (3%) 970 (7%) BAZ <-2 z-scores (%) 26% 21% 23% 11-13.9 years N (%) 359 (3%) 320 (3%) 679 (5%) BAZ <-2 z-scores (%) 40% 36% 38% All N (%) 7172 (51%) 6985 (51%) 14157 (100%) WHZ or BAZ <-2 z-scores (%) 14% 12% 13%

TABLE 33 VALIDITY OF ACTUAL WHO CUT -OFF FOR SEVERE AND ACUTE MALNUTRITION IN CHILDREN <5 Y false age (y) positive sensitivity AUC rate MUAC cut-off 12.5 cm : acute malnutrition boys 0-1.9 y 2.1% 17.8% 0.578 2-4.9 y 0.2% 6.5% 0.531

girls 0-1.9 y 4.9% 32.9% 0.640 2-4.9 y 0.3% 10.2% 0.549

all 0-1.9 y 3.5% 24.3% 0.604 2-4.9 y 0.3% 8.1% 0.539

MUAC cut-off 11.5 cm : severe acute malnutrition boys 0-1.9 y 0.5% 3.3% 0.514 2-4.9 y 0.1% 0.0% 0.500

girls 0-1.9 y 0.9% 18.2% 0.586 2-4.9 y 0.0% 0.8% 0.540

all 0-1.9 y 0.7% 8.6% 0.540 2-4.9 y 0.0% 2.8% 0.514

129

New cut-offs are presented in table 34. When segregated into smaller age groups, sensitivity of MUAC to identify acute malnutrition increased considerably, with sensitivities from 49% to 76% for acute malnutrition, and from 55% to 83% for severe acute malnutrition. Sensitivity of the new cut-offs is higher for acute malnutrition than for severe acute malnutrition.

For new cut-offs, in children above 5y, sensitivity ranged from 58% to 94%. Youden index ranged from 31% to 64% for both acute malnutrition and severe acute malnutritionThe difference between the highest Youden index and the Youden index for the selected cut-off was between 0 and 4%.

The new MUAC cut-offs for acute malnutrition and severe acute malnutrition increase with age (Figure 8) over the whole age range from 0 -14 years. In both children below and above 5 y, cut- offs are generally higher for boys than for girls, but the difference between boys and girls was never more than 0.5cm and both cut-offs followed the same pattern. However, in the age group 8-10.9 y cut-off are higher in girls than in boys. The difference in new MUAC cut-off between identifying acute malnutrition (WHZ<-2) and severe acute malnutrition (WHZ<-3) for each gender and age group was small, and varied from 0.2 cm to 1 cm.

In children below 2 y, MUAC were lower in stunted children than in non-stunted children (table 34). In the children from 2- 5 years, all wasted children were also stunted, so no cut-off for stunted versus non stunted children could be calculated for this age group.

130

TABLE 34 NEW CUT -OFFS * BY AGE GROUP AND GENDER FOR SEVERE AND MODERATE ACUTE MALNUTRITION FOR CHILDREN FROM 0 TO 14 Y

age (y) MUA false specificit sensitiv AUC Youde difference MUAC false specificity sensitivi AUC Youden difference MUAC false specificit sensitivi AUC Youden differe C cut- positiv y ity n with highest cut -off positive ty index with highest cut -off positiv y ty index nce off e rate index Youden rate Youden e rate with index* index* highest Youden index* Boys Girls All Acute malnutrition

All children

0 - 13.9 32% 68% 87% 0.772 54% 0% 13.6 32% 68% 82% 0.746 49% -2% 13.7 30% 70% 81% 0.753 51% 0% 1.9 y 2 - 14.4 30% 70% 82% 0.762 52% -3% 14.2 30% 70% 84% 0.770 54% -1% 14.3 30% 70% 83% 0.762 52% -1% 4.9 y 5 - 15.5 30% 70% 84% 0.769 54% -3% 15.4 33% 67% 85% 0.756 51% -4% 15.4 31% 69% 84% 0.769 54% 0% 7.9 y 8 - 16.4 31% 69% 84% 0.763 53% -1% 16.6 32% 68% 85% 0.764 53% 0% 16.5 32% 68% 83% 0.758 52% -1% 11.9 y 11 - 13.9 18.2 38% 62% 92% 0.768 54% 0% 17.9 21% 79% 84% 0.813 63% -1% 18.2 33% 67% 90% 0.784 57% 0% y

Stunted children

0 - 13.6 31% 69% 85% 0.770 54% -2% 13.2 31% 69% 83% 0.760 52% -1% 13.4 31% 69% 80% 0.74 49% -2% 1.9 y 5 2 - 14 20% 80% 78% 0.792 58% 0% 14 31% 69% 81% 0.750 50% 0% 14 25% 75% 80% 0.77 31% 0% 4.9 y 1

Non stunted children 0 - 14 29% 71% 85% 0.776 55% -1% 13.7 32% 68% 83% 0.755 51% 0% 13.9 34% 66% 85% 0.75 51% 0% 1.9 y 6 2 - N/A : no wasted non stunted children 0% 0% 0% 4.9 y

Severe acute malnutrition

0 - 1.9 y 13.7 29% 71% 82% 0.764 53% -2% 13.4 19% 81% 73% 0.778 55% 0% 13.5 27% 73% 80% 0.76 53% -1% 3 2 - 14.3 31% 69% 83% 0.760 52% -3% 13.8 20% 80% 84% 0.820 64% 0% 14.1 27% 73% 81% 0.76 54% -1% 4.9 y 9 5 - 14.7 11% 89% 58% 0.736 47% -3% 14.7 17% 83% 73% 0.778 56% -4% 14.7 14% 86% 63% 0.74 49% -4% 7.9 y 7 8 - 16 30% 70% 92% 0.807 62% -4% 16.2 31% 69% 91% 0.800 60% -2% 16.1 31% 69% 91% 0.80 60% -3% 11.9 2 y 11 - 13.9 17.3 25% 75% 82% 0.881 56% -4% 17.6 33% 67% 89% 0.779 56% -2% 17.5 31% 69% 89% 0.78 58% 0% y 9 *cut-off selected according to highest AUC -0.02 AND highest sensitivity AND false positive rate ≤33% 131

FIGURE 8 OPTIMAL CUT -OFFS FOR ACUTE MALNUTRITION (AM) AND SEVERE ACUTE MALNUTRITION (SAM) BY AGE GROUP AND GENDER

132

Discussion

In this paper, optimal cut-offs for MUAC to identify acute malnutrition calculated from a pooled dataset of Cambodian children, taking age and gender into account, were higher than the current WHO recommendation. To our knowledge, this is the first paper to report optimal MUAC cut-offs in Asian children, and cut-offs for children >5 y of age. Sensitivity of identifying children with a WHZ <-2 z-score was drastically improved (from <25% to >80%) with the new cut-offs. A consequence of the higher sensitivity is of course a lower specificity, and indeed the number of false-positive cases increased also.

For example, in our dataset, 1045 wasted children (WHZ <-2 z-scores), including 155 children with WHZ <-3 z-score, were not identified as suffering from acute malnutrition by the current WHO MUAC cut-offs of 12.5cm, corresponding to 83% of wasted children below 5 y. Using the new cut-offs for gender and age as recommended in the current paper to screen for acute malnutrition, we missed only 154 wasted children (12.3%). However, using the cut-offs recommended in this paper, 4152 children would have been wrongly identified as having acute malnutrition, while using the current WHO cut-off, only 187 children would have been wrongly identified for acute malnutrition. The difference between new cut-offs for acute and severe acute malnutrition is narrow (table 34) so we suggest to only retain cut-offs for acute malnutrition.

Half a century ago, MUAC was assumed to be almost stable between 0 and 5 years, based on well-nourished Polish children 428 but more recent studies suggested that MUAC is age and gender specific 420,429 . A MUAC for age z-scores reference was built in 1997 but it requires knowledge of the age of the child to the nearest month, something which is often difficult to obtain in the field 295 . Therefore, the present study tried to find a balance between practicality and ideal by dividing children under 5 y in 2 groups using 2 y as a threshold.

The reliability of WHZ as a gold standard to define acute malnutrition is questioned 430 . WHZ is related to body shape and may overestimate the prevalence of acute malnutrition in some populations 431 . Also, it was reported that MUAC performs at least as well or even better than WHZ in predicting mortality among malnourished children 417,432 . However, assessing the immediate death risk is not the only purpose of diagnosing acute malnutrition. Acute malnutrition defined by WHZ<-2 also contributes to increased morbidity, impaired physical and cognitive development, and is associated with micronutrient deficiencies 418 . Therefore acute malnutrition screening should not be reduced to mortality risk screening. Identifying only children at high risk for mortality may prevent to treat children suffering from less advanced acute malnutrition but still at high risk for impaired development. Also, we showed that in the Cambodian context, MUAC and WHZ identified a different sub-set of children as being malnourished. Hence, we have argued that both MUAC and WHZ should be used to identify malnutrition, and one cannot use one or the other 419 .

However, to avoid overburdening the health system with false-positive cases of malnutrition, we suggest distinguishing 2 uses for MUAC cut-off. The first use will be the prevention of death, and the current WHO MUAC cut-off should be used to initiate treatment for severe acute malnutrition in. The newly proposed MUAC cut-offs should be used to screen for children suffering from acute malnutrition. These children may not be at immediate risk for mortality, but early detection of acute malnutrition may prevent them for later risk for mortality and for impaired development.

133

MUAC and WHZ are associated with different aspects of body composition, and therefore identify different groups of children with malnutrition. Therefore, both indicators should be used to start the treatment 419 . Currently, only 1 cut-off for MUAC is being recommended and used for both screening and initiation of treatment for acute malnutrition. Earlier, we have proposed to use 2 different cut-offs for MUAC: one for screening, and one for initiation of treatment for acute malnutrition 419 . The new cut-offs for acute malnutrition proposed in the present paper are meant for screening at community level for acute malnutrition, with treatment for severe acute malnutrition started using WHZ < --3 z-score as indicator of acute malnutrition or the current WHO cut-offs for MUAC. This way, sensitivity of identifying children with acute malnutrition is increased to >80%, whereas specificity for starting treatment remains the same. Using new MUAC cut-off would therefore only increase the number of children being sent to a primary health center to be re-measured for MUAC, weight and height, therefore the global cost of screening malnutrition would increase, as well as more children being correctly treated for malnutrition 429 . As obtaining a MUAC is a non-invasive procedure without any health risks, this poses not a health but an economical issue, as overtreatment can be a burden to the health system. But in return, parents of those children who are likely to be moderately acute malnourished could receive prevention counseling to avoid their children to become severe as no community management of acute malnutrition is being implemented in Cambodia at scale.

Based on the same sample of children under 5 y, Laillou et al recommend to use 13.3 cm as MUAC cut-off to screen severe acute malnutrition in children from 6 m to 5 y 419 , which is lower than the recommended cut-offs in the current paper. This difference can be explained by the fact that in the current paper, the age group was split into 2 (0 - 1.9 y and 2 - 4.9 y). In addition, in the current paper, steps of 0.1 cm were used to identify the new MUAC cut-off, instead of 0.25 cm as in paper by Laillou et al. Finally, we have added the criteria of highest sensitivity and maximum acceptable false positive rate to define the new MUAC cut-off. However, although the cut-offs differ slightly, it is clear that current practice to identify children with acute malnutrition can be improved considerably by increasing the current WHO recommended cut- offs for MUAC.

Children above 5 y should also be taken into account in the management of acute malnutrition. Indeed at school age, acute malnutrition, often accompanied by micronutrient deficiencies, can delay maturation, impair muscular strength, bone density and work capacity 18 . Thus malnutrition at school age increase risk of morbidity, of school failure and school drop-out 17 . MUAC cut-offs gradually increased from birth to adolescence in an almost linear manner, but changes are influenced by changes in growth velocity, e.g. during puberty. In our study, cut-offs were higher for boys than for girls (+0.1 - +0.3 cm), except from age group 8 to 10.9 years. We assume that the earlier adolescent growth spurt in girls than in boys, with associated changes in lean and fat mass, underlies this phenomena 433 . Schools could be a practical platform to follow school-aged children for acute malnutrition.

Height of the child was another important factor in determining the new MUAC cut-offs, with lower values in stunted children compared to non stunted children (table 34). Several studies show that stunted children tend to accumulate more fat mass and gain less lean body mass than non stunted children 434,435 , hence perhaps stunted children have lower muscle mass, leading to lower MUAC cut-offs.

134

A limit of the current study is that data used in this study was from Cambodia and therefore that new MUAC cut-offs presented here are only adapted to Cambodian children. Hence, the aim of the present study is not to suggest using the presented optimal cut-offs as international reference. However, the study highlights that a unique MUAC cut-off for children below 5 y results in a large of children not receiving vital treatment. Therefore we recommend a meta- analysis using the same method on a large dataset from different countries including age, gender, MUAC, height and weight to be conducted, in order to provide an international reference of MUAC cut-offs by gender and age groups to screen for acute malnutrition.

To conclude, new cut-offs for MUAC, adapted to age and gender are needed to improve the sensitivity to identify children with WHZ-score<-2 z-scores. The current WHO MUAC cut-off should be used in targeted interventions and to initiate treatment as it is adapted to detect mortality risk among malnourished children. In the present study, new MUAC cut-offs for screening for acute malnutrition, using age- and gender-specific cut-offs are presented from children from 0 – 14 y. Rapid adaption of these new cut-offs will result in identification of >80% of children with acute malnutrition.

135

5.2. Subclinical inflammation increases plasma transferrin receptor and ferritin and decreases plasma RBP but does not affect plasma zinc in school children and women in Cambodia and Senegal (Manuscript, drafted)

To be submitted Marion Fiorentino a, Marlène Perignon a, Khov Kuong e, Richard de Groot o, Jacques Berger a, Frank Wieringa a

Introduction

Micronutrient deficiencies are a major public health problem, especially in low-income countries. They are widely prevalent in the world with one child over 3 and one pregnant women over 6 being vitamin A deficient, one child or pregnant women over 5 suffering from iron deficiency anemia, and 17% of the world population being at risk for zinc deficiency 3. Micronutrient deficiencies are responsible for increased morbidity and cognitive functions damage. By impairing growth and school achievement in children and affecting reproductive functions and fetal development in women 3, they accentuate the intergenerational cycle of malnutrition 436 .

However, the accuracy of micronutrient deficiencies assessment can be hampered by other factors then micronutrient status affecting biomarkers. Inflammation for example is known to affect biomarkers for iron status (plasma ferritin concentrations) and vitamin A status (plasma retinol concentrations) even in apparently healthy individuals 353 . A complex chain reaction involving immunoregulatory cytokines follows injury or infection and contributes to withholding essential nutrients to infectious agents ('nutritional immunity') and starting repairing process 353 . However, these same mechanisms affect the concentrations of the biomarkers while overall micronutrient status remains the same. At the same time, it appears that micronutrient deficiency can affect the acute phase response 437 . Therefore more precision is needed about the biological response of biomarkers for micronutrient status in relation to inflammation. For biomarkers of vitamin A status (plasma retinol or RBP concentrations) 201,339,438 correction factors based on a meta-analysis have been proposed in order to adjust values in subjects during different stages of inflammation 201 . A similar approach has been used for ferritin concentrations 197 . Iron deficiency is indicated by depleted iron stores , corresponding to low plasma ferritin concentrations (FER) and/or tissue iron deficiency, indicated by elevated soluble transferrin receptor (TFR) concentrations 198 . TFR was long considered to be not or less sensitive to inflammation than FER 198,439 but some studies showed a significant impact of acute phase response on TFR values 440 .

The effect of inflammation on plasma zinc concentration remains unclear too 203,441 . Hence, more data on the impact of acute-phase response on zinc and TFR concentrations is needed. The objective of the present study was to investigate the impact of subclinical inflammation on the assessment of micronutrient deficiencies in 3 different populations and settings (Senegalese school children, Cambodian school children, Cambodian women) by investigating the impact of having elevated acute phase proteins (C-reactive protein and 1-Acidglycoprotein) on biomarkers for iron, vitamin A and zinc status.

136

Participants and methods

Study area, design, population surveyed and ethics

Three datasets were used for the current study.

School children from Senegal

The study was a representative cross-sectional survey conducted in 2010 in children from primary state schools of Dakar, Senegal 225 . A two-stage cluster sampling method was chosen with schools considered as primary sampling units. Within 30 randomly selected schools, and without criteria for age and gender, 20 children were randomized as final sampling units in each school, resulting in a sample of 594 children. The protocol was approved by the ethical committee of the National Health Research of Senegal. The school directors informed parents of the selected children on the purpose and proceedings of the study. Written informed consent was obtained from all parents at the beginning of the study.

School children from Cambodia

Data were collected as part of a randomized placebo-controlled trial investigating the impact of multi-micronutrient fortified rices on health and development of Cambodian schoolchildren (FORISCA UltraRice+NutriRice study, FOrtified RIce for School meals in CAmbodia). Baseline data collection was conducted in November 2012 in 20 primary schools from Kampong Speu province in Cambodia 399 .The schools were randomly selected from primary schools participating in school meal or take-home ration programs of the UN World Food Program (WFP). Children attending the selected schools were eligible to be part of the study if they were between 6 –16 y of age, had written informed consent from parents/caregivers and did not have any mental or severe physical handicap. In each school, 132 children were randomly selected after stratification by sex and grade, hence 2640 children. 169 children were not recruited because they were absent on the day of data collection or refused to participate. Hence, a total of 2471 schoolchildren aged 6 –16 y participated into the blood sample collection. The study was approved by the National Ethic Committee for Health Research (NECHR) of the Ministry of Health, Phnom Penh, Cambodia, the Ministry of Education, Youth and Sports, Phnom Penh, Cambodia, and the Research Ethics Committee of PATH, Seattle, USA.

WRA (Women of Reproductive Age) from Cambodia

The last dataset was provided by a serological survey for antibodies for Tetanus, Rubella and Measles in WRA conducted in Cambodia in 2010 442 . 611 enumeration areas (EAs) were selected from the 28 764 EAs in the 2008 Cambodia General Population Census by probability proportional to size (PPSIn each EA, 22 households were randomly selected. All eligible women in selected households were invited to participate after providing written information about the survey and obtaining consent from women. 2117 women participated into the blood sample collection. We used residual serum from this survey to determine ferritin, sTfR, RBP and acute phase proteins.

137

Blood collection and laboratory analysis

4-5 mL of venous blood was collected by experienced phlebotomists using sterile single-use material: in a vacutainer with heparin (Terumo heparin Venosafe) in Senegal and in Cambodian women, and in a vacutainer with no anticoagulant (Vacuette, Greiner Bio One, AUstria) in Cambodian children, both trace-element free. Blood samples were stored immediately in an icebox containing ice-packs and transported to the laboratory within a maximum of 5 hours after the first sample withdrawal. Plasma/serum samples were separated by centrifugation, aliquoted, and stored at −20°C. Centrifugation resulted in serum samples in Cambodian children and in plasma samples in Cambodian women and Senegalese children. However, we will call plasma all samples in the following text.

For plasma zinc measurement, samples were sent with dry ice, to the Nutripass laboratory of the Institut de Recherche pour le Développement (IRD, Montpellier, France) for Senegalese school children or to the National Institute of Nutrition (Hanoi, Vietnam) for Cambodian school children. Zinc status was not measured in Cambodian WRA. Plasma zinc was measured by flame atomic absorption spectrophotometry (AAS), using trace-elements free procedures and urinary iodine (UIC) was measured using an ammonium persulfate method.

Samples from the 3 (studies were sent with dry ice and to the CBS laboratory (Willstaett, Germany) for determination of retinol-binding protein (RBP), C-reactive protein (CRP), FER, TFR and α1 -acid-glycoprotein (AGP). RBP, FER, TfR, CRP, AGP were measured by a sandwich enzyme-linked immunosorbent assay (ELISA) technique 195 .

Inflammation was determined by elevated CRP (>5 mg/L) and/or elevated AGP (>1 g/L) allowing differentiation between incubation phase (high CRP and normal AGP), convalescence phase (both high AGP and CRP) and late convalescence phase (high AGP and normal CRP) 197 . Iron tissue deficiency was defined by TfR >8.3 mg/L 195 [ and depleted iron stores were defined by FER <15 µg/L [WHO, 2011, Serum ferritin concentrations for the assessment of iron status and iron deficiency in populations. ]. Low FER and high TfR are both considered as indicators of iron deficiency (ID) 198 so ID was defined by iron stores depleted (low FER) and/or iron tissue deficiency (high TfR). Body iron was calculated according to the formula of Cook: body iron (mg/kg) = −(log (TfR/FER ratio)−2.8229)/0.1207. “Body iron deficiency” was defined by body iron <0 199 . TFR/log FER Index was calculated by dividing TFR (mg/L) by log FER (ug/L). 7.06=8.3/log(15) was used to define ID as “high TFR/log TFR index” 443 . Vitamin A status was measured by RBP concentration which reflects plasma retinol concentration because RBP occurs in a 1:1:1 complex with retinol and transthyretin 200 , which is not affected by inflammation 444 . Vitamin A deficiency (VAD) was defined by RBP<0.7 µmol/L 200 and marginal VAD was defined for RBP values ≥0.7 µmol/L and <1.05 µmol/L 202 . Zinc deficiency was defined by plasma zinc <9.9 µmol/L, 10.1 µmol/L or 10.7 µmol/L respectively in children <10 y, girls>10 y and boys >10 y 203 . Children were considered as severely zinc deficient when plasma zinc was <7.7 µmol/L 445 .

Statistical analysis

Data entry, including quality checks and validation by double entry of questionnaires, was performed with EpiData version 3.1 (EpiData, Odense, Denmark). Data management and analyses were performed using SPSS version 20.0 (SPSS, Inc., Chicago, IL). Significance was

138 defined as P<0.05. The distributions of biomarkers concentration were checked for normality using normality plots and Kolmogorov-Smirnoff test. Because distributions of AGP, CRP, TfR, and FER were skewed, they were log transformed. Spearman’s rank correlation co efficients were determined to assess relationships among APP (Acute Phase response Proteins) and micronutrient status biomarkers. For those which are correlated to APP concentrations, we calculated the ratio of the geometric mean or mean values of the biomarker for the group with inflammation, to the reference group without inflammation 197,201,368,440 . The correction factor was calculated as 1/ratio. Prevalence of poor micronutrient status were calculated i) without correction ii) only in the subjects with no inflammation iii) using biomarkers concentrations adjusted with CFs calculated in the present study iv) using FER and RBP concentrations adjusted with CFs recommended by Thurnham 197,201 v) using 30 µg/L as low FER cut-off in children with inflammation 198 . Corrected prevalences were compared to uncorrected prevalence using Mc Nemar’s chi -square.

Results

Biochemical characteristics of the participants from 3 different samples are presented in table 35. Although similar considering gender proportion (half of children were female) and age (10 y), populations of school children from Cambodia and Senegal differed in inflammatory and micronutrient status. Prevalence of inflammation was less than 15% in Senegalese children and Cambodian women but was 40% in Cambodian children, mostly of them being in the late convalescence phase (35% had high AGP and normal CRP). Prevalence of iron deficiency (defined as either a high TFR and/or a low FER) was 39% in Senegalese children, 52% in Cambodian children and 14% in Cambodian women. In Cambodian children, ID was mostly related to a high TFR (only 1% had low FER) while in Senegal children, 33% had a high TFR with 20% having a low FER and in Cambodian women, 10% had a high TFR and 8% had low FER. Prevalence of VAD was less than 5% in all samples but more than 40% of Senegalese children had a marginal vitamin A status. ZD affected 25% of the children in Senegal and more than 90% of the Cambodian children. Correlations between APP and biomarkers of iron, vitamin A and zinc status are presented in table 36. CRP and AGP were highly positively correlated in all samples. FER was highly positively correlated with both CRP and AGP in all samples. TFR was positively correlated to AGP in all samples, and to CRP in the Cambodian school children. Contradictory results were found between indicators based on the TFR/FER ratio (body iron and TFR/log FER index) and APP among the different groups: FER/log FER index was negatively correlated to CRP in both Senegalese children and Cambodian women, while FER/log FER index was positively correlated to AGP in Cambodian children. Similarly, correlations between TBI and AGP were positive in Senegal children and Cambodian women, but negative in Cambodian children.

139

TABLE 35 DEMOGRAPHIC AND BIOCHEMICAL CHARACTERISTICS OF PARTICIPANTS : SCHOOL CHILDREN FROM SENEGAL , SCHOOL CHILDREN FROM CAMBODIA , AND WRA FROM CAMBODIA

Senegalese school Cambodian school Cambodian WRA children (n=594) children (n=2471) (n=2117) Age (y) 10.2 ± 2.4 9.6 ± 2.4 26.3 ± 7.0 Female (%) 52.5 50.3 100.0 CRP (mg/L) 1.2 ± 2.6 1.4 ± 5.0 1.7 ± 4.7 High CRP, >5 mg/L (%) 5.7 5.6 6.8 AGP (g/L) 0.8 ± 0.2 1.0 ± 0.4 0.7 ± 0.2 High AGP, >1 g/L (%) 10.6 39.9 10.3 Inflammation, high CRP or high AGP (%) 12.1 40.3 14.1 No inflammation, normal CRP and normal AGP (%) 87.9 59.7 85.9 Incubation, high CRP and normal AGP (%) 1.5 0.4 3.8 Early convalescence, high CRP and high AGP (%) 4.2 5.3 3 Late convalescence, normal CRP and high AGP (%) 6.4 34.6 7.3 Ferritin (ug/L) 29.7 ± 17.3 88.4 ± 46.4 79.8 ± 58.1 Low ferritin, <12ug/L (%) 20.4 1.3 7.6 TfR (mg/L) 8.1 ± 3.1 8.8 ± 2.7 6.1 ± 2.6 High TfR, >8.3 mg/L (%) 33.3 51.4 10.1 ID (low ferritin and/or high TfR) 38.7 51.6 13.5 Body iron (mg/kg) 2.6 ± 2.9 6.1 ± 2.2 6.8 ± 3.7 ID (Low body iron, <0 mg/kg) (%) 16.7 1.7 5.1 TFR/log FER index 6.5 ± 6.0 4.8 ± 2.2 3.9 ± 3.3 ID (index >7.06) 23.9 7 7.2 RBP (μmol/L ) 1.1 ± 0.3 1.5 ± 0.4 1.8 ± 0.7 Vitamin A deficiency, RBP <0.7 μmol/L (%) 3.7 1.0 0.5 Marginal vitamin A deficiency, RBP <1.05 μmol/L (%) 42.1 11.4 6.0 Zinc (umol/L) 11.4 ± 2.1 7.5 ± 1.9 6.8 ± 3.7 Zinc deficiency * (%) 27.2 92.8 - Severe zinc deficiency ** (%) 4.5 54.2 - In most subjects, RBP was negatively correlated to CRP and to AGP. In Cambodian school children however, RBP was negatively associated with CRP but positively associated with AGP. No correlation between zinc and CRP or AGP was found in subgroups but in all subjects combined, zinc was negatively associated with CRP and AGP. For biomarkers that where significantly correlated to CRP and/or AGP (FER, TFR, RBP and zinc), correction factors were calculated for groups according to inflammatory status (tables 37, 38, 39, 40). Mean FER was significantly higher in participants in incubation, in early convalescence, and in late convalescence, or with elevated AGP or CRP in all subjects (table 37). Effect of inflammation was stronger on FER than on TFR with all CFs being lower than 0.83 and almost all P-values being <0.05. Although the changes in FER between the different phases of inflammation were of the same magnitude between the 3 study populations, there were some significant differences in CFs among the different populations.

140

TABLE 36 SPEARMAN 'S CORRELATION COEFFICIENT (Ρ) BETWEEN INFLAMMATORY AND MICRONUTRIENT STATUS VARIABLES , IN SENEGALESE SCHOOL CHILDREN , CAMBODIAN SCHOOL CHILDREN AND CAMBODIAN WRA

Senegalese Cambodian Cambodian Overall (n=5182) school children school children WRA n(=2117) (n=594) (n=2471)

CRP AGP 0.44 ** 0.58 ** 0.62 ** 0.34 ** FER 0.26 ** 0.27 ** 0.26 ** 0.24 TFR -0.02 0.02 0.13 ** -0.04 TBI 0.16 ** 0.16 ** 0.03 0.16 ** Index -0.12 ** -0.11 * 0.02 -0.13 ** RBP 0.03 * -0.23 ** -0.15 ** 0.14 ** ZN -0.08 ** 0.00 -0.01 -

AGP FER 0.30 ** 0.26 ** 0.35 ** 0.20 TFR 0.40 ** 0.09 * 0.34 ** 0.16 ** TBI -0.01 0.11 * -0.05 * 0.08 ** Index 0.20 ** -0.06 0.17 ** 0.01 RBP -0.01 -0.11 ** 0.11 ** -0.05 * ZN -0.16 ** -0.04 0.01 - * P-value <0.05; ** P-value <0.001

Elevated CRP had no influence on TFR concentrations (table 38), whereas elevated AGP was associated with higher TFR concentrations. TFR concentrations were significantly lower in all subjects combined during the early incubation phase, and higher in later stages of inflammation. . The maximum increase in TFR concentrations in the later stages of inflammation was less than 25% (lowest CF being 0.78). In children, but not in the women, RBP concentrations were significantly lower when CRP was elevated (table 39). During incubation, mean RBP concentrations were lower in the children but higher in women. Concentrations were lower to the same extent in all populations during early convalescence. In all children combined, zinc concentrations tended to be lower in early convalescence and were significantly lower during late convalescence and with elevated AGP (table 40). These effects disappeared in subgroups of Senegalese children and Cambodian children.

Uncorrected prevalence of high TFR was higher than corrected prevalence in all samples but significantly only in Cambodian children and women (table 41). Uncorrected prevalence of low TFR was lower than corrected prevalence, especially in Senegalese children and Cambodian women. Difference between corrected and uncorrected prevalence of ID defined by low FER and/or high TFR was lower than 2% in Senegalese children and Cambodian women, but was around 8% in Cambodian children. Iron deficiency defined by low body iron or high FER/log FER index was less sensitive to inflammation. Prevalence of vitamin A deficiency was below 4% in all samples and difference between adjusted and unadjusted prevalence was below 1%.

141

TABLE 37 CF S AND RATIOS OF FERRITIN BY INFLAMMATORY STATUS IN SENEGALESE SCHOOL CHILDREN , CAMBODIAN SCHOOL CHILDREN AND CAMBODIAN WRA

n (%) FER ( μg/L) 1 P-value 2 Ratio (95%) 3 CF 4 All subjects combined No inflammation 3817 ( 74% ) 53.7 ± 2.2 -- - Incubation 98 ( 2% ) 64.6 ± 2.3 0.023 1.20 ( 1.19 - 1.21 ) 0.83 Early convalescence 219 ( 4% ) 112.2 ± 1.8 0.000 2.09 ( 2.08 - 2.09 ) 0.48 Late convalescence 1048 ( 20% ) 83.2 ± 1.9 0.000 1.55 ( 1.55 - 1.55 ) 0.65 Senegalese school children No inflammation 522(88%) 23.5±1.9 -- - Incubation 9 ( 2% ) 28.5 ± 1.7 0.339 1.22 ( 1.16 - 1.26 ) 0.82 Early convalescence 25 ( 4% ) 46.9 ± 1.7 0.000 2.00 ( 1.96 - 2.03 ) 0.50 Late convalescence 38 ( 6% ) 35.1 ± 1.6 0.000 1.50 ( 1.47 - 1.52 ) 0.67 Cambodian school children No inflammation 1476 ( 60% ) 68.9 ± 1.7 -- - Incubation 9 ( 0% ) 88.9 ± 1.7 0.141 1.29 ( 1.27 - 1.31 ) 0.77 Early convalescence 130 ( 5% ) 125.8 ± 1.6 0.000 1.83 ( 1.82 - 1.83 ) 0.55 Late convalescence 856 ( 35% ) 87.5 ± 1.7 0.000 1.27 ( 1.27 - 1.27 ) 0.79 Cambodian WRA No inflammation 1819 ( 86% ) 56.4 ± 2.3 -- - Incubation 80 ( 4% ) 67.9 ± 2.4 0.051 1.20 ( 1.19 - 1.21 ) 0.83 Early convalescence 64 ( 3% ) 120.4 ± 1.7 0.000 2.14 ( 2.13 - 2.14 ) 0.47 Late convalescence 154 ( 7% ) 73.8 ± 2.4 0.000 1.31 ( 1.30 - 1.32 ) 0.76

All subjects combined Normal AGP 3915 ( 76% ) 54.3 ± 2.2 -- - High AGP 1267 ( 24% ) 86.9 ± 1.9 0.000 1.60 ( 1.60 - 1.6 ) 0.62 Senegalese school children Normal AGP 531 ( 89% ) 23.5 ± 1.9 -- - High AGP 63 ( 11% ) 38.9 ± 1.7 0.000 1.65 ( 1.63 - 1.68 ) 0.60 Cambodian school children Normal AGP 1485 ( 60% ) 69.0 ± 1.7 -- - High AGP 986 ( 40% ) 91.2 ± 1.7 0.000 1.32 ( 1.32 - 1.32 ) 0.76 Cambodian WRA Normal AGP 1899 ( 90% ) 56.8 ± 2.3 -- - High AGP 218 ( 10% ) 85.1 ± 2.3 0.000 1.50 ( 1.49 - 1.50 ) 0.67

All subjects combined Normal CRP 4865 ( 94% ) 59.2 ± 2.2 -- - High CRP 317 ( 6% ) 93.7 ± 2.0 0.000 1.58 1.58 1.59 0.63 Senegalese school children Normal CRP 560 ( 94% ) 24.1 ± 1.9 -- - High CRP 34 ( 6% ) 41.1 ± 1.8 0.000 1.71 ( 1.68 - 1.73 ) 0.59 Cambodian school children Normal CRP 2332 ( 94% ) 74.1 ± 1.0 -- - High CRP 139 ( 6% ) 123.0 ± 1.6 0.000 1.66 ( 1.66 - 1.66 ) 0.60 Cambodian WRA Normal CRP 1973 ( 93% ) 57.6 ± 2.3 -- - High CRP 144 ( 7% ) 87.1 ± 2.2 0.000 1.51 ( 1.51 - 1.52 ) 0.66

1 Geometric means ± SD 2 ANOVA on log-transformed FER means of positive group VS control group 3 Ratio of back-transformed FER concentrations of positive group VS control group 4 CF=1/Ratio

142

TABLE 38 CF S AND RATIOS OF TRANSFERRIN RECEPTOR BY INFLAMMATORY STATUS IN SENEGALESE SCHOOL CHILDREN , CAMBODIAN SCHOOL CHILDREN AND CAMBODIAN WRA

n (%) TFR (mg/L) 1 P-value 2 Ratio (95%) 3 CF 4 All subjects combined No inflammation 3817(74%) 6.8±1.4 -- - Incubation 98 ( 2% ) 5.8 ± 1.3 0.000 0.85 ( 0.81 - 0.89 ) 1.17 Early convalescence 219 ( 4% ) 7.8 ± 1.4 0.000 1.16 ( 1.12 - 1.18 ) 0.87 Late convalescence 1048 ( 20% ) 8.7 ± 1.3 0.000 1.29 ( 1.27 - 1.29 ) 0.78 Senegalese school children No inflammation 522 88%) 7.7±1.3 -- - Incubation 9 2% ) 7.5 ± 7.7 0.746 0.97 ( 0.32 - 1.63 ) 1.03 Early convalescence 25 4% ) 8.2 ± 1.3 0.266 1.07 ( 1.00 - 1.13 ) 0.94 Late convalescence 38 6% ) 7.9 ± 1.2 0.653 1.02 ( 0.97 - 1.08 ) 0.98 Cambodian school children No inflammation 1476(60%) 8.0±1.3 -- - Incubation 9 ( 0% ) 7.2 ± 1.3 0.206 0.90 ( 0.79 - 1.01 ) 1.12 Early convalescence 130 ( 5% ) 8.8 ± 1.3 0.000 1.10 ( 1.07 - 1.13 ) 0.91 Late convalescence 856 ( 35% ) 9.3 ± 1.3 0.000 1.15 ( 1.15 - 1.18 ) 0.87 Cambodian WRA No inflammation 1819(86%) 5.7±1.4 -- - Incubation 80 ( 4% ) 5.5 ± 1.3 0.296 0.96 ( 0.91 - 1.02 ) 1.04 Early convalescence 64 ( 3% ) 6.0 ± 1.4 0.145 1.06 ( 0.99 - 1.11 ) 0.94 Late convalescence 154 ( 7% ) 6.4 ± 1.4 0.000 1.13 ( 1.08 - 1.16 ) 0.89

All subjects combined Normal AGP 3915 ( 76% ) 6.8 ± 1.4 -- - High AGP 1267 ( 24% ) 8.6 ± 1.4 0.000 1.27 ( 1.25 - 1.28 ) 0.79 Senegalese school children Normal AGP 531 ( 89% ) 7.7 ± 1.3 -- - High AGP 63 ( 11% ) 8.0 ± 1.3 0.301 1.04 ( 0.99 - 1.08 ) 0.96 Cambodian school children Normal AGP 1485 ( 60% ) 8.0 ± 1.3 -- - High AGP 986 ( 40% ) 9.2 ± 1.3 0.000 1.15 ( 1.14 - 1.16 ) 0.87 Cambodian WRA Normal AGP 1899 ( 90% ) 5.7 ± 1.4 -- - High AGP 218 ( 10% ) 6.3 ± 1.4 0.000 1.11 ( 1.07 - 1.14 ) 0.90

All subjects combined Normal CRP 4865 ( 94% ) 7.2 ± 1.4 -- - High CRP 317 ( 6% ) 7.1 ± 1.4 0.847 1.00 0.96 - 1.01 ) 1.00 Senegalese school children Normal CRP 560 ( 94% ) 7.7 ± 1.3 -- - High CRP 34 ( 6% ) 8.0 ± 1.3 0.449 1.04 ( 0.98 - 1.10 ) 0.96 Cambodian school children Normal CRP 2332 ( 94% ) 8.5 ± 1.3 -- - High CRP 139 ( 6% ) 8.7 ± 1.3 0.238 1.03 ( 1.00 - 1.05 ) 0.97 Cambodian WRA Normal CRP 1973 ( 93% ) 5.7 ± 1.4 -- - High CRP 144 ( 7% ) 5.7 ± 1.4 0.170 1.00 ( 0.96 - 1.04 ) 1.00

1 Geometric means ± SD 2 ANOVA on log-transformed TFR means of positive group VS control group 3 Ratio of back-transformed TFR concentrations of positive group VS control group 4 CF=1/Ratio

143

TABLE 39 CF S AND RATIOS OF RBP BY INFLAMMATORY STATUS IN SENEGALESE SCHOOL CHILDREN , CAMBODIAN SCHOOL CHILDREN AND CAMBODIAN WRA

n (%) RBP (μmol/L ) P-value 2 Ratio (95%) 3 CF 4 All subjects combined No inflammation 3817 74% 1.58 ± 0.6 -- - Incubation 98 2% 1.80 ± 0.7 0.000 1.14 ( 1.05 - 1.23 ) 0.8775 Early convalescence 219 4% 1.31 ± 0.4 0.000 0.83 ( 0.79 - 0.86 ) 1.2097 Late convalescence 1048 20% 1.58 ± 0.5 0.996 1.00 ( 0.98 - 1.02 ) 1.00 Senegalese school children No inflammation 522(88%) 1.13± 0.3 -- - Incubation 9 ( 2% ) 0.86 ± 0.2 0.002 0.76 ( 0.64 - 0.88 ) 1.31 Early convalescence 25 ( 4% ) 0.97 ± 0.3 0.003 0.86 ( 0.75 - 0.97 ) 1.16 Late convalescence 38 ( 6% ) 1.06 ± 0.3 0.092 0.94 ( 0.85 - 1.03 ) 1.07 Cambodian school children No inflammation 1476 ( 60% ) 1.48 ± 0.4 -- - Incubation 9 ( 0% ) 1.26 ± 0.5 0.090 0.85 ( 0.63 - 1.07 ) 1.17 Early convalescence 130 ( 5% ) 1.26 ± 0.3 0.000 0.85 ( 0.82 - 0.89 ) 1.17 Late convalescence 856 ( 35% ) 1.56 ± 0.4 0.000 1.05 ( 1.03 - 1.08 ) 0.95 Cambodian WRA No inflammation 1819 ( 86% ) 1.79 ± 0.7 -- - Incubation 80 ( 4% ) 1.96 ± 0.6 0.015 1.09 ( 1.02 - 1.17 ) 0.91 Early convalescence 64 ( 3% ) 1.53 ± 0.6 0.002 0.85 ( 0.77 - 0.94 ) 1.17 Late convalescence 154 ( 7% ) 1.80 ± 0.7 0.736 1.01 ( 0.94 - 1.07 ) 0.99

All subjects combined Normal AGP 3915 ( 76% ) 1.58 ± 0.6 -- - High AGP 1267 ( 24% ) 1.53 ± 0.5 0.003 0.97 ( 0.93 - 0.95 ) 1.03 Senegalese school children Normal AGP 531 ( 89% ) 1.13 ± 0.3 -- - High AGP 63 ( 11% ) 1.03 ± 0.3 0.003 0.91 ( 0.84 - 0.98 ) 1.10 Cambodian school children Normal AGP 1485 ( 60% ) 1.48 ± 0.4 -- - High AGP 986 ( 40% ) 1.52 ± 0.4 0.014 1.03 ( 1.01 - 1.05 ) 0.97 Cambodian WRA Normal AGP 1899 ( 90% ) 1.79 ± 0.7 -- - High AGP 218 ( 10% ) 1.72 ± 0.6 0.136 0.96 ( 0.91 - 1.01 ) 1.04

All subjects combined Normal CRP 4865 ( 94% ) 1.58 ± 0.5 -- - High CRP 317 ( 6% ) 1.46 ± 0.6 0.000 0.92 ( 0.90 - 0.98 ) 1.08 Senegalese school children Normal CRP 560 ( 94% ) 1.13 ± 0.3 -- - High CRP 34 ( 6% ) 0.94 ± 0.3 0.000 0.83 ( 0.74 - 0.92 ) 1.20 Cambodian school children Normal CRP 2332 ( 94% ) 1.51 ± 0.4 -- - High CRP 139 ( 6% ) 1.28 ± 0.4 0.000 0.85 ( 0.80 - 0.89 ) 1.18 Cambodian WRA Normal CRP 1973 ( 93% ) 1.79 ± 0.7 -- - High CRP 144 ( 7% ) 1.77 ± 0.7 0.804 0.99 ( 0.92 - 1.06 ) 1.01

1 Geometric means ± SD 2 ANOVA on log-transformed RBP means of positive group VS control group 3 Ratio of back-transformed RBP concentrations of positive group VS control group 4 CF=1/Ratio

144

TABLE 40 CF S AND RATIOS OF ZINC BY INFLAMMATORY STATUS IN SENEGALESE SCHOOL CHILDREN , CAMBODIAN SCHOOL CHILDREN AND CAMBODIAN WRA

Zinc ( μmol /L) n (%) 2 3 4 1 P-value Ratio (95%) CF All children combined No inflammation 1716 ( 67% ) 8.41 ± 1.3 -- - Incubation 15 ( 1% ) 9.18 ± 1.4 0.219 1.09 ( 0.98 - 1.11 ) 0.92 Early convalescence 120 ( 5% ) 8.00 ± 1.3 0.053 0.95 ( 0.85 - 1.00 ) 1.05 Late convalescence 698 ( 27% ) 7.70 ± 1.3 0.000 0.91 ( 0.95 - 0.97 ) 1.09 Senegalese school children No inflammation 1203 ( 88% ) 11.19 ± 7.4 -- - Incubation 7 ( 2% ) 11.72 ± 1.1 0.512 1.05 ( 0.96 - 1.08 ) 0.95 Early convalescence 95 ( 4% ) 10.76 ± 1.3 0.340 0.96 ( 0.95 - 1.02 ) 1.04 Late convalescence 662 ( 6% ) 11.34 ± 1.2 0.703 1.01 ( 0.98 - 1.03 ) 0.99 Cambodian school children No inflammation 513 ( 60% ) 7.45 ± 1.2 -- - Incubation 8 ( 0% ) 6.93 ± 1.2 0.399 0.93 ( 0.88 - 1.05 ) 1.07 Early convalescence 25 ( 5% ) 7.41 ± 1.3 0.795 1.00 ( 0.98 - 1.02 ) 1.00 Late convalescence 36 ( 35% ) 7.53 ± 1.3 0.278 1.01 ( 1.00 - 1.02 ) 0.99

All children combined Normal AGP 1731 ( 68% ) 8.42 ± 1.3 -- - High AGP 818 ( 32% ) 7.74 ± 1.3 0.000 0.92 ( 0.91 - 0.93 ) 1.09 Senegalese school children Normal AGP 521 ( 90% ) 11.20 ± 1.2 -- - High AGP 61 ( 10% ) 11.10 ± 1.2 0.740 0.99 ( 0.98 - 1.02 ) 1.01 Cambodian school children Normal AGP 1210 ( 62% ) 7.45 ± 1.2 -- - High AGP 757 ( 38% ) 7.52 ± 1.3 0.339 1.01 ( 1.00 - 1.01 ) 0.99

All children combined Normal CRP 135 ( 5% ) 8.2 ± 1.3 -- - High CRP 2414 ( 95% ) 8.1 ± 1.3 0.712 0.99 0.96 1.02 1.01 Senegalese school children Normal CRP 549 ( 94% ) 11.2 ± 1.2 -- - High CRP 33 ( 6% ) 11.0 ± 1.2 0.593 0.98 ( 0.98 - 1.06 ) 1.02 Cambodian school children Normal CRP 1865 ( 95% ) 7.5 ± 1.2 - - High CRP 102 ( 5% ) 7.4 ± 1.3 0.514 0.99 ( 0.98 - 1.05 ) 1.01

1 Geometric means ± SD 2 ANOVA on log-transformed zinc means of positive group VS control group 3 Ratio of back-transformed zinc concentrations of positive group VS control group 4 CF=1/Ratio

145

TABLE 41 EFFECT OF CORRECTING TFR, PF, RBP CONCENTRATIONS ON THE PREVALENCE OF LOW IRON STATUS AND LOW VITAMIN A STATUS , IN SCHOOL CHILDREN FROM CAMBODIA , SCHOOL CHILDREN FROM SENEGAL AND WRA FROM CAMBODIA

Senegalese school Cambodian school Cambodian WRA children (n=594) children (n=2471) (n=2117) % P-value* % P-value* % P-value* Low Fer 1 uncorrected 20.4 NA 1.3 NA 7.6 NA in participants with no inflammation 22.6 NA 1.4 NA 8.0 NA corrected for 3 phases of inflammation 2 21.5 0.016 1.4 0.250 7.9 0.008 corrected for 3 phases of inflammation (Thurnham) 3 22.9 0.000 1.8 0.000 8.0 0.002 FER < 30 ug/L in children with inflammation 23.2 0.000 2.1 0.000 8.9 0.000

High TfR 4 uncorrected 33.3 NA 51.4 NA 10.1 NA in participants with no inflammation 32.6 NA 42.2 NA 9.5 NA corrected for 3 phases of inflammation 5 33.2 1.000 42.9 0.000 9.7 0.021

Iron deficiency (high TfR and/or low Fer) 1 4 uncorrected 38.7 NA 51.6 NA 13.5 NA in participants with no inflammation 38.5 NA 42.5 NA 13.2 NA corrected for 3 phases of inflammation 2 5 37.9 0.180 44.5 0.000 13.4 0.791 corrected for 3 phases of inflammation (Thurnham) 3 39.1 0.500 51.6 1.000 13.8 0.016 FER < 30 ug/L in children with inflammation 40.2 0.004 51.7 0.500 14.3 0.000

Iron deficiency (low body iron 6 ) uncorrected 16.7 NA 1.7 NA 5.1 NA in participants with no inflammation 18.6 NA 1.7 NA 5.3 NA corrected for 3 phases of inflammation 2 5 17.5 0.002 1.8 0.002 5.2 0.002 corrected for 3 phases of inflammation (Thurnham) 3 17.7 0.000 1.9 0.000 5.3 0.000

Iron deficiency (high TFR/logFER index 7) uncorrected 23.9 NA 7.0 NA 7.2 NA in participants with no inflammation 25.7 NA 6.3 NA 7.1 NA corrected for 3 phases of inflammation 2 5 24.6 0.125 6.3 0.000 7.2 1.000 corrected for 3 phases of inflammation (Thurnham) 3 24.6 0.125 7.7 0.000 7.3 0.500 FER < 30 ug/L for children with inflammation 8 26.9 0.000 13.2 0.000 7.9 0.000

146

TABLE 41 (CONTINUED ) EFFECT OF CORRECTING TFR, FER, RBP CONCENTRATIONS ON THE PREVALENCE OF LOW IRON STATUS AND LOW VITAMIN A STATUS , IN SCHOOL CHILDREN FROM CAMBODIA , SCHOOL CHILDREN FROM SENEGAL AND WRA FROM CAMBODIA

Senegalese school Cambodian school Cambodian WRA children (n=594) children (n=2471) (n=2117) % P-value* % P-value* % P-value* Vitamin A deficiency 9 uncorrected 3.7 NA 1.0 NA 0.5 NA in participants with no inflammation 2.9 NA 0.9 NA 0.4 NA corrected for 3 phases of inflammation 10 3.4 0.500 0.9 1.000 0.4 0.500 corrected for 3 phases of inflammation (Thurnham) 11 3.0 0.125 0.7 0.031 0.3 0.250

Marginal vitamin A status 12 uncorrected 42.1 NA 11.4 NA 6.0 NA in participants with no inflammation 39.7 NA 11.0 NA 5.7 NA corrected for 3 phases of inflammation 10 39.6 0.000 11.7 0.427 5.9 0.687 corrected for 3 phases of inflammation (Thurnham) 11 38.9 0.000 8.7 0.000 5.6 0.008

Zinc deficiency 13 uncorrected 27.2 NA 92.8 NA NA NA in participants with no inflammation 27.5 NA 93.3 NA NA NA corrected for 3 phases of inflammation 5 27.3 1.000 92.8 0.500 NA NA

1 FER < 15 ug/L for children and WRA, except if notified 2 cf. CFs table 3 3 CFs Thurnham for incubation, early convalescence, late convalescence : 0.64 0.39 0.65 in children; 0.73 0.58 0.85 in women 4 TFR> 8.3 mg/L 5 cf. CFs table 4 6 body iron < 0 mg/kg 7 TFR/logFER> 7.05 (corresponding to 8.3/log15) 8 TFR/logFER> 5.3 (corresponding to 8.3/log30) 9 RBP<0.7 µmol/L 10 cf. CFs table 5 11 CFs Thurnham for incubation, early convalescence, late convalescence : RBP 0.87 0.76 0.89 12 RBP<1.05 µmol/L NA : Non applicable * McNemar's chi-square of proportion to compare uncorrected prevalence and corrected prevalence

147

Discussion

Animal models have shown an effect of inflammation on plasma zinc concentrations. The effect of the APR on plasma zinc concentrations is related to the redistribution of zinc from the plasma to the liver via the activation of immunoregulatory cytokines. For example, interleukin 6 stimulates the production of metallothionin and of zinc transport Zip14, which regulates hepatic zinc uptake and plasma zinc concentration 203,441,446,447 . In animal models, the fall of plasma zinc concentrations occurs before clinical symptoms (hence just after infection) and return to normal quickly 441 . In our cohorts in school children in Cambodia and Senegal, late phases of inflammation slightly affected plasma concentrations of zinc but the effect was modest. The prevalence of zinc deficiency after adjustment was similar to the prevalence among children without inflammation (respectively 27.3% and 27.5% in Senegalese children and 92.8% and 92.8% in Cambodian school children). Duncan et al 448 reported zinc concentrations of Scottish hospital patients being significantly negatively correlated to CRP (ρ= -0.16). In these hospital patients, the degree of inflammation was much more severe than in our normal populations, and therefore cannot be classified as 'sub-clinical inflammation'. Indeed, in the study of Duncan et al, zinc concentrations were not significantly lower in patients with CRP concentrations <20 g/L, a situation similar to ours. Therefore our findings suggest that for determining zinc status in normal populations that is in population with only a low prevalence of sub-clinical infection, taking the acute phase response into account will not lead to a significant improvement in the estimate of the prevalence of zinc deficiency. This is consistent with the conclusions of an earlier review indicating no effect of inflammation on zinc status in children 441 but in contrast to the recommendations of the IZINCG working group.

Plasma RBP concentrations are often used as biomarker for vitamin A status, and consider equal as plasma retinol concentrations as retinol:RBP ratio is not altered by inflammation 444 . During the APR proinflammatory cytokines inhibit the hepatic synthesis of RBP and secretion of RBP:retinol complex 438 which decrease retinol mobilization. Currently, it is not clear whether the resultings hyporetinemia is a primary goal of the APR or a secondary consequence 449 .

In our cohorts of children from both Senegal and Cambodia RBP was decreased during incubation and early convalescence, which is consistent with other studies 201,339,438 . Indeed, retinol is known to fall rapidly within 48 hours 353 . During late convalescence, RBP was still lower in Senegalese children but higher in Cambodian children compared to participants with no inflammation , as has been reported earlier too, in a small sample of Indonesian school children 450 . A meta-analysis using the same CRP and AGP cut-offs showed that the APR-induced decrease of retinol lasted into late convalescence 201 . But in this meta-analysis populations with a poor vitamin A status were used (almost all median retinol < 1.05 µmol/L).Retinol (and hence RBP) concentrations are expected to rebound during convalescence 449 . We believe that baseline vitamin A status might explain this difference between Senegalese and Cambodian children. Vitamin A status in the Senegalese children was poor (40% marginal vitamin A status), whereas in Cambodian children, vitamin A status was good. This means that when correcting for inflammation, baseline vitamin A status should be taken into account. More data on populations with a relative good vitamin A status is needed to confirm this. But if found correct, it will affect the way to correct for inflammation for vitamin A status, as shown by the differences in vitamin A prevalence after using different correction factors. In Senegalese children, the correction factors as proposed by Thurnham (and calculated using populations

148 with poor vitamin A status) resulted in a prevalence of vitamin A deficiency very close to the prevalence in children without inflammation. In Cambodian subjects however, the correction factors proposed by Thurnham lead to an underestimation of the prevalence of vitamin A deficiency, presumably because vitamin A concentrations rebounded earlier in this population.

In all populations, FER was highly positively correlated with APP as reported earlier 368,440 . In addition, TFR was positively correlated to AGP, and also to CRP in Cambodian school children. But TFR concentrations were only increased in groups with elevated AGP,This is consistent with another study suggesting that FER reacts faster to the APR than TFR 440 . Again, cytokines are involved in the modulation of iron metabolism during the acute phase response: TNF, possibly via stimulation by IFN- γ or IFN -α may induce ferritin synthesis 451,452 . No change of TFR concentrations according to inflammatory status was found in Senegalese children, which could be linked to the low prevalence of inflammation. Although TFR has been long considered as more reliable than FER to assess ID when inflammation was prevalent 198 , in Cambodian women and children, inflammation significantly increased TFR concentrations as it was found in other studies 440,453 ,. Oppositely, and in line with Righetti et al 440 , a significant decrease of TFR concentrations was observed in all subjects combined during incubation .

Not adjusting ferritin concentrations for inflammation leads to underestimate the proportion of low iron stores, especially in populations with endemic inflammation. In our cohorts adjustment for inflammation increased the prevalence of low ferritin by 2% in Cambodian school children (prevalence of inflammation 40%), by 1.1% in Senegalese children, and by just 0.3% in Cambodian women. These small changes in adjusted prevalence are readily explained by either the low prevalence of inflammation in Cambodian women and Senegalese children and by the low prevalence of low iron stores in Cambodian children and women. Overall, using 30 µg/L as a cut-off for children with inflammation performed less efficiently than adjustment for inflammation by use of APP. To our knowledge, no correction factors for TFR concentration for inflammation have been published. In our cohorts, TFR concentrations were higher during inflammation leading to an overestimation of iron deficiency defined by tissue iron deficiency, in line with results from Righetti et al 440 . In the cohort of Cambodian children, who had a high prevalence of inflammation, the prevalence of high TFR was almost 10% higher as compared to children without inflammation (51% vs 42%). Similar adjustment on TFR were used in Ivorian infants with similarly high proportion of inflammation 440 . In Cambodian women and Senegalese children where inflammation was not very prevalent, the difference between adjusted and not adjusted prevalence of tissue iron deficiency did not exceed 1%.

The use of TFR/log FER rather than TFR/FER to predict iron stores deficiency has been recommended 454 , especially in areas with endemic infection 443 . Instead of calculating CFs to apply to FER/log FER index as it was done by Righetti et al 440 , we used corrected values of FER and TFR to calculate FER/log FER index or body iron and prevalence of ID defined by FER/log FER index > 7.06 or by negative body iron. Depending on the iron status of the population, that is, whether iron deficiency was primarily indicated by low FER or by high TFR, inflammation resulted in either over- or underestimation of the prevalence of ID defined by high FER/log FER index or negative body iron. Overall, the difference between uncorrected prevalence and the reference prevalence was low (< 2%), suggesting that indicators based on both TFR and FER are relatively insensitive to inflammation. The adjusted prevalence of iron deficiency using our CFs for both FER and TFR or Thurnham CFs for FER only were was very close in Cambodian women and Senegalese children, suggesting that in these cohorts adjusting FER concentrations for

149 inflammation is more important than adjusting TFR for inflammation to obtain a more precise estimate of ID. However, using Thunham Cfs for FER only in the Cambodian children led to overestimation of ID, even more than without adjusting for inflammation, because in this cohort, FER concentrations were high, making TFR the most important factor in these 2 indicators.

Prevalence of inflammation is considered to be low under 15% and to only have a modest impact on the assessment of micronutrient deficiencies 439,455 . However consensus about cut- offs to use to define elevated APP is needed. Recently, the sensitivity of the now commonly used cut off for CRP <5 mg/L has been questioned 353 . Several studies used different than 5 mg/L for CRP and 1 g/L for AGP or a series of cut-offs to define categories of inflammation 339,456,457 . Some research showed similar patterns of impact of acute phase response on micronutrient biomarkers using CRP cut-off either 5 mg/L or 10 mg/L 458 while other indicated that the best CRP cut-off could vary between 5, 10 or 20 mg/L depending on the measured biomarker 448 . Different cut-offs across class ages may be needed 339 . Hence, an international recommendation about APP cut-offs used to categorize inflammation is needed.

Our study had several limitations. First, the iron status of populations was heterogeneous with Cambodian children having adequate iron stores but a high proportion of high TFR, while ID in Cambodian women and in Senegalese children was both due to low iron stores and iron tissue deficiency.. But it gave us the opportunity to examine the impact of inflammation on micronutrient status in variable population profiles. Second, the prevalence of inflammation in Senegalese children and Cambodian women was low. Third the size of sample of Senegalese children was small compared to Cambodian children and women. Consequently, the impact of inflammation on some biomarkers in some inflammatory subgroups may have been blurred. Fourth, the golden standard of measuring iron deficiency is bone marrow 371 , which is invasive, expensive and not ethical to perform for research purposes only, so that we did not collect it in our studies.

Conclusion

Our findings suggest to use TFR/log FER to assess populations iron status and iron deficiency prevalence if CRP and AGP are not possible to measure. Indeed, among various indicators of iron deficiency (low FER, high TFR, low FER and/or high TFR, negative body iron, high FER/log FER index), FER/log FER index is the less sensitive to inflammation. Considering the sensitivity of FER and TFR examined alone, they should be adjusted for inflammation preferably by both CRP and AGP , especially in areas with a high infection pressure. So far, only few studies, including the present one, calculated correction factors for TFR 440 . The impact of inflammation on TFR concentrations is clear, as well as the need to adjust TFR concentrations for inflammation in order to avoid the overestimation of ID defined by tissue iron deficiency. However, a large meta-analysis or pooled analyses such as carried out by Thurnham et al on retinol and ferritin 197,201 is needed to quantify better the impact of inflammation in different stages on TFR. The impact of inflammation on zinc status has been found modest in the present study, indicating that plasma zinc concentrations are perhaps less affected by inflammation than previously thought. Different responses of retinol to incubation and convalescence phases were observed in the present study, presumably depending on the vitamin A status of the population. More research should be conducted on APP cut-offs to define the different phases of inflammation, and the sensitivity of micronutrient biomarkers.

150

General conclusion

Education is crucial to break the cycle of poverty and malnutrition 379,380 . Nutritional deficiencies are particularly harmful for the physical and cognitive development of school-aged children and adolescents who have high nutrient requirements to support their skeletal and brain growth spurt. The associations between micronutrient deficiencies, parasite infection and poor cognitive performance observed in our research in Cambodia highlight the importance of addressing health and nutrition issues in school-aged children and adolescents in an holistic way, in order to support the educational achievement of the future generation. We showed that some of the adverse effects of malnutrition are still reversible at school age, with cognitive performance being improved by one type of multiple-micronutrient fortified rice. Poor dietary habits and high rates of acute malnutrition and micronutrient deficiencies as observed in the Senegalese urban school children also underline the need for school-based interventions. When entering school, children spend more time out-door, away from their family supervision and consume street and processed foods in urban areas when no school feeding is available 145 , which makes the food transition between preschool and school age a sensitive health issue. Therefore nutrition education and school feeding should be extended, especially in disadvantaged urban areas. To convince policy makers on the need to scale-up of school feeding programs, strong scientific evidence of its benefits on health and nutrition is needed. Our research showed promising beneficial effects of fortified rice on micronutrient status and cognitive performance after only 6 months. A larger impact may be expected from long-term daily consumption of fortified rice. In the future, effectiveness studies of fortified food conducted over more than one school year may stimulate the introduction of fortified staple food in existing school feeding programs.

Furthermore, more research on indicators is needed to better characterize malnutrition in school-aged children and adolescents, taking into account intra-populations variability, effects of environmental factors and specific needs of this age group. We established cut-offs for MUAC in Cambodia, which is a cost-effective tool to screen for acute malnutrition among school children. And we suggested correction factors to eliminate the cofounding impact of the acute- phase response on TFR concentrations, a biomarker of iron status, and verified other correction factors of ferritin and vitamin A status.

More research and efforts should be shifted to the nutritional status of adolescent girls. Nutritional deficiencies among still growing adolescents, who are potential young mothers, may be particularly harmful for both mother and infant, reinforcing the intergenerational cycle of malnutrition. Specific nutrient requirements have to be established for pregnant adolescent girls. School-based and community-based interventions should be implemented to prevent precocious marriage and to improve the nutrition of adolescent girls.

Neglecting the issue of malnutrition in school-aged children and adolescents may compromise the benefits of the historical improvement achieved in younger children. Investing in nutrition in this age range when resources are available is an opportunity to optimize the development of next generations in developing countries.

151

Authors affiliation

a UMR Nutripass, Institut de Recherche pour le développement, Montpellier, France b GRET, Dakar, Senegal c Division du Contrôle Médical Scolaire, Ministère de l’Enseignement, Dakar, Senegal d Department of Statistics Iowa state University, Ames, United States of America e Department of Fisheries Post-Harvest Technologies and Quality Control, Ministry of Agriculture, Forestry and Fisheries, Phnom Penh, Cambodia f United Nations World Food Programme, Phnom Penh, Cambodia g PATH, Seattle, United States of America h Department of Psychology, Royal University of Phnom Penh, Phnom Penh, Cambodia i Department of Health Sciences, VU University Amsterdam, Amsterdam, The Netherlands j Department of Biomedical Sciences, Institute of Tropical Medicine, Antwerp, Belgium k Laboratory of Pathophysiology, University of Antwerp, Wilrijk, Belgium l National Institute of Hygiene, Epidemiology and Microbiology, Havana, Cuba m Pedro Kourí Institute of Tropical Medicine, Havana, Cuba. n Department of Nutrition, Exercise and Sports, Copenhagen University, Frederiksberg, Denmark o Independant consultant, Phnom Penh, Cambodia. p National Nutrition Program, Maternal and Child Health Center, Phnom Penh, Cambodia. q UNICEF, Maternal Child Health and Nutrition section, Phnom Penh, Cambodia.

152

Synthesis

Background Malnutrition remains a crucial public health problem, especially among young children where it is responsible for growth retardation in one third of them and for half of deaths, mostly in developing countries. Nutritional deficiencies during fetal life and early childhood have long- term and possibly irreversible adverse effects on physical and cognitive development. Consequently, most of the efforts have been focused on improving nutrition among women of reproductive age and children under 5 y. However, prevalence of chronic and acute malnutrition and micronutrient deficiencies are estimated to be also high in older children and adolescents, especially in Africa and in South-East Asia.

Africa South-East Asia Stunting 22% 29% Thinness 35% 35% Iron deficiency 29% 20% Vitamin A deficiency 32% 17% Iodine deficiency 33%, 33%, Zinc deficiency 54% 29% Estimation of prevalence of nutritional deficiencies in school-aged children (Best, 2007)

The growth spurt in adolescence is the second most important in life after the one occurring in the first year of life, which dramatically increases energy, protein and micronutrients nutrient requirements. Although growth retardation reflects primarily nutritional deprivation during the first ‘1000 days’ period, improving nutrition during late childhood and adolescence can prevent the continuation of the stunting process, which is known to be associated with impaired work capacity and obstetric complications. Due to the maternal-fetal nutrient competition in pregnant adolescents supporting the triple burden of poor dietary intake, their own growth and their fetus growth, adolescent mothers and their infants are particularly at risk for nutritional deficiencies in developing countries, where early marriage and childbearing are common. School age and adolescence are also critical periods for neurological development, with frontal lobes achieving development spurts at 7-9 years and in the mid-teenage years. Research shows beneficial effects of iron or iodine interventions on cognitive performance of school children. Improving micronutrient status in school-aged children reduces morbidity and enhances school performance. Therefore, nutrition at school age and adolescence is crucial for educational achievement and future adult life. However, data about nutritional status and impact of interventions in school-aged children and adolescents are scarce. The objective of the thesis was to evaluate the prevalence of malnutrition, its determinants and consequences on health outcomes among school-aged children and adolescents from two developing countries, as well as and the effectiveness of school-based interventions. The thesis is based on 2 studies conducted among African and South-East-Asian school-aged children and adolescents.

153

The first study evaluated the nutritional status and nutrient intake of children from primary state schools of Dakar, the capital of Senegal. Indeed, the urban population has been rapidly expanding in the developing world, and tremendous discrepancies within urban have arisen, which was aggravated by the food price crisis and the global crisis in the past decade. Plus, urban lifestyle and transition from infancy to school age expose children to poor dietary and health habits. However, still few data are available about the nutritional status of school-aged children and adolescents living in urban areas.

The second study is an effectiveness study of micronutrient-fortified rice distributed through the School Feeding Program of World Food Program (WFP) in the Kampong Speu province, Cambodia. School feeding has been scaled up in the past decades because of its incentive effect on school enrollment and attendance but only 18% of school children in low-income countries receive school meals and the impact on nutritional status remains small compared to the costs. Rice being massively consumed in many Asian and Sub-Saharan countries, distributing fortified rice through school feeding may be a cost effective way to improve micronutrient status of school-aged children and adolescents.

Methods

· Studies design

In Senegal, a cross-sectional survey was conducted in 2010 on a representative sample of ~ 600 children aged 5-17 y from primary state schools of the Dakar region (Senegal), selected through a two-stage random cluster sample (30 schools × 20 children) and enrolled after written informed parental consent.

In Cambodia, the FORISCA-UltraRice+NutriRice study was a cluster-randomized, double- blinded, placebo-controlled trial conducted in 2012 among ~2400 children aged 6-16 y in the Kampong Speu province in 2012. Sixteen schools participating in the School Feeding Program of WFP were randomly assigned to receive fortified rice (UltraRice®original, UltraRice®new or NutriRice®) or unfortified rice (placebo) 6 days per week for 6 months. Within each school, 132 children were randomly selected with stratification on grade and gender, and enrolled in the study after written informed parental consent resulting in ~2400 children. Data were collected before the intervention (baseline), after 3 months (midline) and after 6 months at the end of the intervention (endline).

Micronutrient composition of uncooked rice per 100g of blended rice Vitamin Vitamin Vitamin Vitamin Vitamin Iron Zinc Vitamin B6 B12 B1 B9 B3 (mg) (mg) A (IU) (mg) (µg) (mg) (mg) (mg) UltraRice®original 10.7 3.0 0.0 0.0 0.0 1.1 0.2 0.0 UltraRice®new 7.6 2.0 2140.0 0.0 3.8 1.4 0.3 12.6 NutriRice® 7.5 3.7 960.0 0.9 1.3 0.7 0.1 8.0

154

· Data collection Once in Senegal, and at baseline, midline and endline in Cambodia, anthropometric data (height, weight, MUAC, triceps skinfold) and blood and urine samples were collected to measure hemoglobin, plasma concentrations of ferritin (FER), soluble transferrin receptors (TFR), retinol binding protein and zinc, and urinary iodine concentration. Correction factors were used for FER and RBP in subjects with inflammation determined with C-reactive protein (CRP) a nd α - 1-acid-glycoprotein (AGP).

In Senegal, two quantitative 24-hours food recalls were conducted among children and quantitative recipes were collected among mothers in households and in street vendors.

In Cambodia, data on cognitive performance (picture completion and block design from WISC III and Raven’s colored progressive matrices) were collected at baseline, midline and endline, as well as feces samples to measure parasite infestation.

155

Prevalence of nutritional deficiencies in the studied populations

Definition Senegal Cambodia Mean age (y) 9.7 10.2 (Age range) (5-17) (6-16) Stunting (%) Height-for-Age z-scores (HAZ) 4.9 40.0 <-2 Severe stunting (%) HAZ <-3 0.7 10.9 Thinness (%) BMI-for-Age z-scores (BAZ) <-2 18.6 25.6 Severe thinness (%) BAZ <-3 5.6 5.0 Overweight (%) BAZ > 1 0.3 0.1 Anemia (%) Hemoglobin (Hb) < 115 g/L 14.4 15.7 (children <12 y); <120 g/L (children 12 - 15 y and girls ≥15 y); <130 g/L for (boys ≥15 y) Severe anemia (%) Hb <70 g/L 0.1 0.1 Depleted iron stores (%) FER corrected for inflammation 21.4 1.5 <15 µg/L Tissue iron deficiency (%) TFR >8.3 mg/L 33.3 51.0 Iron deficiency (ID) (%) depleted iron stores and/or 39.1 51.2 tissue iron deficiency Iron deficiency Anemia (IDA) (%) ID and anemia 10.6 10.0 Vitamin A deficiency (%) RBP corrected for inflammation 3.0 0.7 <0.7 µmol/L Marginal vitamin A status (%) corrected RBP<1.05 µmol/L 35.9 7.9 Zinc deficiency (%) zinc <0.65 mg/L (<10 y); <0.70 25.9 92.8 mg/L (boys≥10 y); <0.66 mg/L (girls ≥10y) Iodine deficiency (%) urinary iodine <100 µg/L 32.8 17.3 Inflammation (%) CRP >5 mg/L and/or AGP >1 12.0 39.5 g/L Parasite infection (%) ≥ 1 egg /g feces - 18.0

156

Risk factors for poor nutrition and health outcomes

Data from Senegal and baseline data from Cambodia were used to evaluate determinants and adverse effects of nutritional deficiencies among school children.

· Nutrient intake and risk factors for micronutrient deficiencies in Senegalese children

Energy and nutrient intakes were adjusted for intra individual variability to obtain estimated usual intakes and prevalence of inadequate intake, using the software PC-Side and according to the intakes recommendations of the Institute of Medicine.

Most of children had insufficient energy intake. 7% had insufficient intake of proteins per kg of body weight. Fiber intake was insufficient in almost 90% of the children. Contribution of proteins and carbohydrates to energy intake was insufficient in respectively 31% and 5% of children. No children had excessive contribution of protein and carbohydrates to energy intake, but respectively 32%, 21% and 14% of children had excessive contribution of lipids, Saturated Fatty Acids (SFAs), and Poly-Unsaturated Fatty Acids (PUFAs) to total energy intake. Half of the children had insufficient iron or vitamin C intake, more than two third had insufficient zinc or vitamin A intake and all of the children had insufficient calcium and folic acid intake. Risks for iron and zinc deficiency were higher when intakes were insufficient in iron, zinc, and protein and excessive in lipid. No micronutrient was consumed above its respective Upper Limits.

These findings indicate a diet poor in dairy products, meat, fruit and vegetables, with a special concern about zinc, vitamin A, folic acid and calcium. Inadequacy of nutrient intake increased with age and is likely to be the main cause of high rates of acute malnutrition and micronutrient deficiencies in the population of primary state schools of Dakar. Reduction of familial care, increase of street food consumption, conjectural food prices crisis, and absence of school meals 145 may lead to a poorer diet at school age compared to infancy, which seem to be aggravated at adolescence in our study. These findings highlight the need of nutritional interventions in Senegalese urban schools such as school feeding program and nutrition education.

· Risk factors for low cognitive scores and impaired growth in Cambodian children

Multivariate analyses were performed to evaluate the associations between cognitive tests scores and variables assessing nutritional status (stunting, anemia and micronutrient status) while taking into account the effect of covariates (age, gender, parasite infection, socioeconomic status). Linear regression analysis was performed with height for age z-scores, plasma zinc and parasite infection with inflammation, age and sex as covariates.

Stunted children had cognitive scores significantly lower than non-stunted children on all tests. In Raven’s colored progressive matrices test, boys with iron -deficiency anemia had lower scores than boys with normal iron status (-1.46, p<0.05). In picture completion test, children with normal iron status tended to score higher than iron-deficient children with anemia (-0.81; p<0.1) or without anemia (-0.49; p<0.1). Parasite infection was associated with an increase in risk of scoring below the median value in block design test (OR=1.62; p<0.05), and with lower scores in other tests, for girls only (both p<0.05). Plasma zinc was positively associated with

157 height for age z-scores (aB=-0.033, p<0.05), but parasite infection was not. Parasite infection showed a negative association with zinc concentration (aB=-0.233, p = 0.05).

Poor cognitive performance of Cambodian school-children was multifactorial and significantly associated with long-term (stunting) and recent nutritional status indicators (iron status), as well as parasite infection. Our findings also suggest that factors influencing child growth may depend on prevalence of parasite species and zinc deficiency. Further research is needed to elucidate these relationships and their underlying mechanisms. A life-cycle approach with programs to improve nutrition in early life and at school-age could contribute to optimal cognitive performance.

Effectiveness of multiple-micronutrient fortified rice on micronutrient status and cognitive performance through the school feeding program of WFP in Cambodia

Generalized mixed models (linear or binary logistic regression) adjusted for age and gender were used to evaluate the impact of intervention groups on micronutrient status and on the scores of the cognitive tests. Secondary analysis were performed including interactions with low iron status, parasite infection and stunting and the impact of intervention, as .these variables were earlier identified as risk factors for low cognitive scores.

After 6 months, children receiving Nutririce® (NR) and Ultrarice®new (URN) respectively had 4 and 5 times less risk of marginal vitamin A status (respectively OR=0.24, p<0.001 and OR=0.20, p<0.001) in comparison to the unfortified rice group. Hemoglobin significantly increased (+0.8g/L, p<0.05) after 3 months in the URN group in comparison to the unfortified rice group, however, after 6 months, this difference was significant only in children with inflammation (+2.1 g/L p<0.01). All cognitive scores improved after 6 months (p<0.001). Block design score improvement was significantly higher in children consuming UltraRice®Original (p<0.05) compared to the other fortified rice groups and the unfortified rice group. No difference among groups was found on Raven’s colored progressive matrices scores or picture completion scores. Stunting, parasite infestation and inflammation negatively affected the impact of the intervention.

Multi-micronutrient fortified rice containing vitamin A effectively improved the vitamin A status of schoolchildren. Impact on hemoglobin and iron status was limited, partly by sub-clinical inflammation. However, fortified rice with the highest concentration of iron improved cognitive performance. Using existing school feeding programs to distribute fortified rice to school children may be a cost-effective way to improve micronutrient status and cognitive performance and thereby improve increase school performance and educational achievements.

158

Indicators to evaluate malnutrition among school-aged children and adolescents

Data from Senegal and baseline data from Cambodia were used to evaluate the relevance of some indicators of malnutrition among school children.

· MUAC cut-offs to screen for acute malnutrition in Cambodian school-aged children and adolescents

As seen in our studies, acute malnutrition is widely spread in school-aged children and adolescents. The latest WHO guidelines for the treatment of severe acute malnutrition recommend establishing MUAC cut-offs for children >5 y. Indeed, even though it is less linked to mortality compared to younger children, acute malnutrition is also important to address during school years as it can impair physical and mental development in this age group.

Anthropometric data from the Cambodian dataset were used. To assess the performance of MUAC cut-offs compared to the golden standard recommended by WHO to define severe and moderate acute malnutrition (BAZ <-2 and BAZ <-3), receiver operating characteristic curves (ROC curves) were constructed. MUAC cut-offs with the highest sensitivity and a false positive rate below 33% were selected among cut-offs with the highest values of area-under-curves (AUC) to be the optimal cut-offs to screen for acute malnutrition.

MUAC optimal cut-offs for acute malnutrition (AM) and severe acute malnutrition (SAM) by age group and gender

MUAC cut-offs gradually increased with age in an almost linear manner, but changes are influenced by changes in growth velocity, e.g. during puberty. Cut-offs were higher for boys than for girls (+0.1 +0.3 cm), except from age group 8 to 10.9 years. We assume that the earlier

159 adolescent growth spurt in girls than in boys, with associated changes in lean and fat mass, underlies this phenomenon.

· Impact of subclinical inflammation on plasma transferrin receptor, ferritin, RBP and zinc in school children in Cambodia and Senegal

Acute-phase response is known to affect plasma ferritin and retinol while micronutrient status remain the same, leading to respectively underestimate and overestimate iron and vitamin A deficiencies. TFR was long considered to be not or less sensitive to inflammation than ferritin but recent research suggest a significant impact of acute phase response on TFR values. The effect of inflammation on plasma zinc concentration remains unclear too. We used data from Cambodia and Senegal to evaluate the impact of subclinical inflammation on the assessment of micronutrient deficiencies among school children.

The ratio of the mean values of the biomarker for the group with inflammation to the reference group without inflammation was calculated. The correction factor was calculated as 1/ratio and applied to calculate prevalence of micronutrient deficiencies adjusted for inflammation.

160

Ratios and correction factors of ferritin, transferrin receptor, RBP and zinc for inflammation status Ferritin Transferrin receptor Ratio (95%) 1 CF 2 Ratio (95%) 1 CF 2 Senegalese school children Incubation 1.22 [1.16-1.26] 0.82 0.97 [0.32-1.63] 1.03 Early convalescence 2.00 [1.96-2.03] 0.50 1.07 [1.00-1.13] 0.94 Late convalescence 1.50 [1.47-1.52] 0.67 1.02 [0.97-1.08] 0.98 Cambodian school children Incubation 1.29 [1.27-1.31] 0.77 0.90 [0.79-1.01] 1.12 Early convalescence 1.83 [1.82-1.83] 0.55 1.10 [1.07-1.13] 0.91 Late convalescence 1.27 [1.27-1.27] 0.79 1.15 [1.15-1.18] 0.87 RBP Zinc Ratio (95%) 1 CF 2 Ratio (95%) 1 CF 2 Senegalese school children Incubation 0.76 [0.64-0.88] 1.31 1.05 [0.96-1.08] 0.95 Early convalescence 0.86 [0.75-0.97] 1.16 0.96 [0.95-1.02] 1.04 Late convalescence 0.94 [0.85-1.03] 1.07 1.01 [0.98-1.03] 0.99 Cambodian school children Incubation 0.85 [0.63-1.07] 1.17 0.93 [0.88-1.05] 1.07 Early convalescence 0.85 [0.82-0.89] 1.17 1.00 [0.98-1.02] 1.00 Late convalescence 1.05 [1.03-1.08] 0.95 1.01 [1.00-1.02] 0.99 1 Ratio of back-transformed concentrations of positive group vs. control group 2 CF=1/Ratio

Secondly, prevalence were adjusted using correction factors for RBP and FER suggested by Thurnham et al.

Senegalese children Cambodian children

Zinc deficiencyZD Zinc deficiency

Marginal VA status * Marginal VA status Marginal VA* status *

VAD Vitamin A deficiency VAD

ID (high index) * * ID (high index) ID (high index)

ID (low body iron) * ID (low body iron) * ID (low body iron) *

ID (High TFR / low FER) * * ID (highTfR/lowFER) ID (highTfR/lowFER) * * Low FER * Low FER Low FER * * * *

High TfR High HighTFR TfR *

0 5 10 15 20 25 30 35 40 45 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 Prevalence (%) Prevalence (%) FER cut-off 30 µg/L used for children with inflammation Uncorrected Corrected with Thurnham CFs Corrected with our CFs (3 phases of inflammation) No inflammation Uncorrected * P-value<0.05 Mc Nemar test to compare prevalence with uncorrected prevalence : Z D: Zinc deficiency; VAD : Vitamin A deficiency; ID : Iron deficiency Effect of correcting TFR, FER, RBP and zinc for inflammation on the prevalence of micronutrient deficiencies

161

In previous research, transferrin receptor was thought to be more reliable than ferritin in endemic inflammation areas, but we showed that transferrin receptor is also sensitive to acute- phase response and that it should be adjusted. Our analysis demonstrated that inflammation increases ferritine and transferrin receptor, leading to underestimate low ferritin and overestimate high transferrin receptor so the index transferrin receptor/log ferritin was the less sensitive indicator to inflammation. RBP decreased during inflammation leading to overestimate vitamin A deficiency or marginal vitamin A status in populations with elevated inflammation rates. Our findings suggest that for determining zinc status in normal populations that is in population with only a low prevalence of sub-clinical infection, taking the acute phase response into account will not lead to a significant improvement in the estimate of the prevalence of zinc deficiency. Overall, more research is needed about the indicators of nutritional status among school-aged children and adolescents. The intra-populations variability, the effect of environmental factors and the specific needs of this age range for each gender should be accounted.

Conclusion

Positive associations between micronutrient deficiencies, parasite infection and poor cognitive performance were observed in Cambodia, underlining the need to address health and nutrition issues among school-aged children and adolescents. Although often omitted by public research, it is known that high nutrient requirements associated to the growth spurt and cerebral development make school-aged children and adolescents particularly vulnerable to poor nutrition, which compromises school achievement, crucial to break the intergenerational cycle of poverty and malnutrition. Vitamin A status and cognitive performance was improved by multiple micronutrient fortified rice in our study in Cambodia, showing that some adverse effects of malnutrition can still be reversed at school age.

Poor dietary intake and elevated rates of acute malnutrition and micronutrient deficiencies in urban school children which do not have access to school feeding as observed in Senegal highlight the need of school-based interventions. Nutrition education and school feeding should be extended, especially in disadvantaged urban areas.

To convince policy makers on the need to scale up school feeding programs, a strong scientific evidence of its beneficial effects on health and nutrition is needed. Promising benefits on micronutrient status and cognitive performance were showed after only 6 months of fortified rice consumption in our research. A larger impact may be expected from long-term daily consumption of fortified rice. In the future, effectiveness studies conducted over more than one school year may stimulate the introduction of fortified staple food in existing school feeding programs.

Moreover, more research on indicators taking into account intra-populations variability, influence of environmental factors and specific needs for school-aged children and adolescents is needed to better identify and determine malnutrition in this age group. We established cut- offs for MUAC in Cambodia, which is a cost-effective tool to screen for acute malnutrition among school children. We suggested correction factors to eliminate the cofounding impact of the acute-phase response on TFR plasma concentrations, a biomarker of iron status, and verified correction factors of ferritin and vitamin A status.

162

Neglecting the issue of malnutrition in school-aged children and adolescents may compromise the benefits of the historical improvement achieved in younger children. Investing in nutrition in these populations is an opportunity to optimize the development of next generations in developing countries.

163

Synthèse

Contexte

La malnutrition est toujours un grave problème de santé publique, notamment parmi les jeunes enfants chez qui elle est responsable du retard de croissance d’un tiers d’entre eux, et de la moitié des décès, majoritairement dans les pays en développement. Les carences nutritionnelles pendant la vie fœtale et la petite enfance causent des dommages à long terme, parfois irréversibles, sur le développement physique et cognitif. Par conséquent, la plupart des efforts ont été concentrés à ce jour sur l’amélioratio n de la nutrition des femmes en âge de procréer et des enfants âgés de moins de 5 ans. Cependant, on estime que la prévalence de la malnutrition chronique et aigue et les carences en micronutriments est également élevée parmi les enfants plus âgés et les adolescents, notamment en Afrique et en Asie du Sud-Est.

Afrique Asie du Sud-Est

Retard de croissance 22% 29%

Maigreur 35% 35%

Carence en fer 29% 20%

Carence en vitamine A 32% 17%

Carence en iode 33%, 33%,

Carence en zinc 54% 29%

Estimation de la prévalence des carences nutritionnelles chez les enfants d’âge scolaire (Best, 2007)

Le pic de croissance pendant l’adolescence est le second plus important après celui qui se déroule pendant la première année de vie, ce qui augmente considérablement les besoins en énergie, protéines et micronutriments. Bien que le retard de croissance reflète principalement une privation nutritionnelle pendant la période des premiers « 1000 jours », améliorer la nutrition pendant l’âge scolaire et l’adolescence peut préve nir le prolongement du processus de retard de croissance, dont on sait qu’il est associé à une capacité de travail réduite et à des complications obstétriques. En raison de la compétition nutritionnelle materno-foetal chez les adolescentes enceintes suppor tant la triple charge d’un régime alimentaire déficient, de leur propre croissance et de la croissance de leur fœtus, les mères adolescentes et leurs nourrissons sont particulièrement à risque de souffrir de carences nutritionnelles dans les pays en développement, ou le mariage et la grossesse précoces sont répandus. L’âge scolaire et l’adolescence sont aussi des périodes critiques pour le développement neurologique, les lobes frontaux connaissant des pics de développement à 7-9 ans et au milieu de l’adolescence. Des recherches mettent en évidence des effets bénéfiques d’interventions en fer ou en iode sur les performances cognitives d’enfants d’âge scolaire. Corriger le statut en micronutriments d’enfants d’âge scolaire réduit la morbidité et améliore les performances

164 scolaires. Ainsi, une nutrition adéquate à l’âge scolaire et à l’adolescence est cruciale pour la réussite scolaire et la vie future. Cependant, les données sur l’état nutritionnel et l’impact d’interventions chez les enfants d’âge scolaire e t les adolescents sont rares. L’objectif de cette thèse est d’évaluer la prévalence de différentes formes de malnutrition, ses facteurs déterminants et ses conséquences sur la santé chez les enfants d’âge scolaire et les adolescents, ainsi que l’efficacité d’interventions scolaires. Cette thèse est basée sur deux études conduites sur des enfants d’âge scolaire et des adolescents de deux pays en développement (Sénégal et Cambodge).

La première étude évalue le statut nutritionnel et les ingérés en nutriments d’enfants des écoles primaires publiques de Dakar, la capitale du Sénégal. En effet, la population urbaine a connu une expansion rapide dans les pays en développement, et des inégalités massives au sein des zones urbaines sont apparues, ce qui s’est encor e aggravé durant la dernière décennie en raison de la crise des prix alimentaires et de la crise économique mondiale. En outre, le mode de vie urbain et la transition de la petite enfance à l’âge scolaire exposent les enfants à de mauvaises habitudes alimentaires et sanitaires. Cependant, très peu de données sont actuellement disponibles sur l’état nutritionnel des enfants d’âge scolaire et les adolescents vivant en zones urbaines. La deuxième étude est une étude d’efficacité de riz fortifié en micronutrim ents distribué via le programme d’alimentation scolaire du Programme Alimentaire Mondial (PAM) dans la province de Kampong Speu au Cambodge. L’alimentation scolaire a connu une certaine expansion dans les dernières décennies en raison de son effet incitatif bénéfique sur les inscriptions et la fréquentation scolaires, mais seulement 18% des enfants d’âge scolaire des pays à faible revenu en bénéficient, et l’impact sur l’état nutritionnel demeure modeste par rapport aux couts engendrés. Le riz étant massivement consommé dans beaucoup de pays asiatiques et sub- sahariens, distribuer du riz fortifié via les circuits d’alimentation scolaire pourrait être une stratégie à haut rapport cout/efficacité d’améliorer le statut en micronutriments des enfants d’âge scola ire et des adolescents.

165

Méthodes

· Plan des etudes

Au Sénégal, une étude transversale fut conduite en 2010 sur un échantillon représentatif d’environ 600 enfants âgés de 5 à 17 ans des écoles primaires publiques de la région de Dakar (Sénégal), sélectionnés par un échantillonnage aléatoire en grappes (30 écoles x 20 enfants).

Au Cambodge, l’étude FORISCA -UltraRice+NutriRice conduite en 2012 sur environ 2400 enfants âgés de 6 à 16 ans dans la province de Kampong Speu était un essai contrôlé randomisé en grappes en double aveugle. 16 écoles participant au programme d’alimentation scolaire du PAM furent réparties de façon aléatoire pour recevoir du riz fortifié (UltraRice®original, UltraRice®new or NutriRice®) ou non fortifié 6 jours par semaine pendant 6 mois. Au sein de chaque école, 132 enfants furent sélectionnés de façon aléatoire en stratifiant par classe et par sexe et recrutés après signature du consentement éclairé par les parents, résultant en un échantillon d’environ 2400 enfants. Les données furent collectées avant l’intervention (baseline), aprés 3 mois (midline), et après 6 mois à la fin de l’intervention (endline).

Composition en micronutriments du riz cru pour 100 grammes de riz mélangé Fer Zinc Vitamine Vitamine Vitamine Vitamine Vitamine Vitamine (mg) (mg) A (IU) B6 (mg) B12 (µg) B1 (mg) B9 (mg) B3 (mg) UltraRice®original 10.7 3.0 0.0 0.0 0.0 1.1 0.2 0.0 Ultr aRice®new 7.6 2.0 2140.0 0.0 3.8 1.4 0.3 12.6 NutriRice® 7.5 3.7 960.0 0.9 1.3 0.7 0.1 8.0

· Collecte des données

Une seule fois au Sénégal, et lors des baseline, midline et endline au Cambodge, des données anthropométriques (taille, poids, périmètre brachial (PB), pli cutané tricipital), ainsi que des échantillons de sang et d’urine furent collectés pour mesurer le taux d’hémoglobine, les concentrations plasmatiques en ferritine (FER), récepteur soluble de la transferrine (TFR), « retinol binding protein » (RBP), et en zinc, et la concentration urinaire en iode. Les concentrations en FER et RBP furent corrigés pour le statut inflammatoire déterminé avec la protéine C-réactive (CRP) et l’ α-1-glycoprotéine acide (AGP).

Au Sénégal, deux rappels alimentaires des 24 heures quantitatifs furent conduits auprès des enfants et les recettes quantitatives furent collectées auprès des mères à domicile et auprès des vendeuses de rues.

Au Cambodge, des données sur les performances cognitives (« picture completion » et « block design » du WISC III et les matrices colorées progressives de Raven) furent collectées lors des baseline, midline and endline, ainsi que des échantillons de fèces pour mesurer l’infestation parasitaire.

166

Prévalence des carences nutritionnelles dans les populations étudiées

Cambodia Definition Senegal

Age moyen (ans) 9.7 10.2 (Tranche d’âge ) (5-17) (6-16)

Retard de croissance (%) z-scores Taille-pour-Age (TAZ) <-2 4.9 40.0

Retard de croissance sévère TAZ <-3 0.7 10.9 (%) Maigreur (%) z-scores IMC-for-Age (IAZ) <-2 18.6 25.6

Maigreur sévère (%) IAZ <-3 5.6 5.0

Surpoids (%) IAZ > 1 0.3 0.1

Anémie (%) Hemoglobine (Hb) < 115 g/L enfants <12 14.4 15.7 ans; <120 g/L enfants 12 - 15 ans et filles ≥15 ans; <130 g/L garçons ≥15 ans

Anémie sévère (%) Hb <70 g/L 0.1 0.1

Stock de fer insuffisant (%) FER corrigée pour inflammation <15 µg/L 21.4 1.5

Carence en fer des tissus (%) TFR >8.3 mg/L 33.3 51.0

Carence en fer (%) carence en fer des tissus et/ou stock de fer 39.1 51.2 insuffisant Anémie ferriprive (%) carence en fer et anémie 10.6 10.0

Carence en vitamine A (%) RBP corrigé pour inflammation <0.7 3.0 0.7 µmol/L Statut marginal en vitamine A RBP corrigé pour inflammation <1.05 35.9 7.9 (%) µmol/L Carence en zinc (%) zinc <0.65 mg/L (enfants <10 ans); <0.70 25.9 92.8 mg/L (garçons ≥10 ans); <0.66 mg/L (filles ≥10 ans) Carence en iode (%) iode urinaire <100 µg/L 32.8 17.3

Inflammation (%) CRP >5 mg/L et/ou AGP >1 g/L 12.0 39.5

Infection parasitaire (%) ≥ 1 oeuf /g fèces - 18.0

Facteurs de risque pour de faibles « outcomes » nutrition et santé

Les données du Sénégal et de la baseline au Cambodge furent analysées pour évaluer les facteurs déterminants et les répercussions négatives de la malnutrition chez les enfants d’âge scolaire.

167

· Apports en nutriments et facteurs de risque de carences en micronutriments

Les apports en énergie et en nutriments furent ajustés en fonction de la variabilité intra- individuelle pour obtenir les « apports usuels estimés” et la prévalence des ingérés inadequats, en utilisant le logiciel PC-Side et en suivant les recommandations d e l’Institut de Médecine.

La plupart des enfants eurent des apports en énergie insuffisants. 7% eurent des apports insuffisants en protéine par kg de poids corporel. Les apports en fibres furent insuffisants chez près de 90% des enfants. La contribution des protéines et des glucides à l’apport énergétique global était insuffisante dans respectivement 31% et 5% des enfants. Aucun enfant n’eut une contribution excessive des protéines et des glucides à l’apport énergétique global, mais respectivement 32%, 21% et 14% des enfants eurent une contribution excessive des lipides, des acides gras saturés (AGS) et des acides polyinsaturés (AGPI) à l’apport énergétique. La moitié des enfants eurent des apports insuffisants en fer et en vitamine C, plus de deux tiers eurent des apports insuffisants en zinc et en vitamine A, et tous les enfants eurent des apports insuffisants en calcium et en acide folique. Les risques de carence en fer et en zinc étaient plus élevés quand les apports étaient insuffisants en fer, zinc, protéines ou excessifs en lipides. Aucun micronutriment ne fut consommé au-dessus de ses limites maximales.

Ces résultats indiquent un régime pauvre en produits laitiers, viande, fruits et légumes, et notamment des apports en zinc, vitamine A, acide folique et calcium particulièrement préoccupants. L’insuffisance des apports augmenta avec l’âge et est probablement la cause principale des taux élevés de malnutrition aigüe et de carences en micronutriments que nous avons observés au sein de la population des écoles primaires publiques de Dakar. La diminution des soins familiaux, l’augmentation de la consommation de nourriture de rue, la crise conjoncturelle des prix alimentaires et l’absence de cantines scolaires peuvent être à l’origine d’un régime alimentaire plus pauvre à l’âge scolaire que pendant la petite enfance, ce qui semble s’aggraver à l’adolescence. Ces résultats mettent en évidence le besoin d’interventions nutritionnelles chez les enfants Sénégalais urbains d’âge scolaire, tels que les cantines scolaires et l’éducation à la nutrition.

· Facteurs de risque pour de faibles scores cognitifs et croissance ralentie chez les enfants cambodgiens

Des analyses multivariées furent effectuées pour évaluer les associations entre les tests cognitifs et les indicateurs de statut nutritionnel (retard de croissance, anémie et statuts en micronutriments) tout en prenant en compte l’effet des covariables (âge, sexe, infection parasitaire, statut socioéconomique). Des régressions linéaires avec les z-scores taille pour âge, concentration plasmatique de zinc et infection parasitaire furent effectuées, avec âge, sexe et inflammation comme covariables.

Les enfants en retard de croissance eurent des scores cognitifs significativement plus bas que les enfants en crois sance normale pour tous les tests. Les garçons souffrant d’anémie ferriprive eurent des scores au test des matrices de Raven inferieurs aux garçons sans carence en fer (- 1.46, p<0.05). Les scores « picture completion » des enfants sans carence en fer eurent tendance à être plus élevés que ceux des enfants carencés en fer (-0.81; p<0.1) et souffrant d’anémie ou sans anémie (-0.49; p<0.1). L’infection parasitaire fut associée à une augmentation

168 du risque d’avoir un score inferieur à la médiane pour le test « block design » (OR=1.62; p<0.05), et à des scores plus faibles dans les autres tests, chez les filles (p<0.05 pour les deux tests). La concentration plasmatique de zinc (aB=-0.033, p<0.05) mais pas l’infection parasitaire, fut positivement associée avec les z-scores Taille-pour-Age L’infection parasitaire fut négativement associée avec la, concentration plasmatique (aB=-0.233, p = 0.05).

Les faibles performances cognitives chez les enfants cambodgiens d’âge scolaire furent multifactorielles et significa tivement associées avec des indicateurs d’historique nutritionnel (retard de croissance) et de statut nutritionnel récent (statut en fer), et avec l’infection parasitaire. Nos résultats suggèrent également que la croissance des enfants pourrait dépendre de l’infection parasitaire et de la carence en zinc. Une recherche plus approfondie est nécessaire pour élucider ces associations et leurs mécanismes sous-jacents. Une approche englobant des programmes d’amélioration de la nutrition pendant la petite enfance et l’âge scolaire pourrait contribuer à optimiser les performances cognitives.

Efficacité du riz fortifié en micronutriments sur les statuts en micronutriments et les performances cognitives via le programme d’alimentation scolaire du PAM au Cambodge

L’impact de l’intervention sur les statuts en micronutriments et les scores cognitifs fut évalué à l’aide de modèles mixtes généralisés (régressions linéaires ou logistiques binaires). Des analyses secondaires incluant les interactions de l’intervention avec le statut en fer, l’infection parasitaire et le retard de croissance furent effectuées, ces variables ayant été identifiées au préalable comme facteurs de risque pour de faibles scores cognitifs.

Après 6 mois, les enfants recevant NutriRice® (NR) et Ultrarice®New (URN) eurent respectivement 4 et 5 fois moins de risque d’avoir un statut marginal en vitamine A (respectivement OR=0.24, p<0.001 and OR=0.20, p<0.001) par rapport à ceux recevant du riz non fortifié. L’hémoglobine diminua significativement (+ 0.8g/L, p<0.05) après 3 mois dans le groupe URN par rapport au groupe « riz non fortifié », cependant, après 6 mois, cette différence était significative seulement chez les enfants sans inflammation (+2.1 g/L p<0.01). Tous les scores cognitives augmentèren t après 6 mois (p<0.001). L’augmentation des scores au test “block design” fut significativement plus élevée chez les enfants recevant Ultrarice®Original (p<0.05) que chez ceux recevant les autres riz fortifiés et le riz non fortifié. Aucune différence ne fut observée entre les groupes pour les tests des matrices de Raven ou « picture completion ». Le retard de croissance, l’infection parasitaire et l’inflammation eurent un effet négatif sur l’impact de l’intervention.

Le riz fortifié en micronutriments contenant de la vitamine A améliora le statut en vitamine A des enfants. L’impact sur l’hémoglobine et le statut en fer fut limité, en partie par l’inflammation sub-clinique. Cependant, le riz fortifié contenant le plus fort taux de fer améliora les performa nces cognitives. Utiliser les programmes existant d’alimentation scolaire peut être une stratégie rentable d’amélioration des statuts en micronutriments et les performances cognitives et ainsi de favoriser la réussite scolaire.

169

Indicateurs d’évaluation de la malnutrition chez les enfants d’âge scolaire et les adolescents

Comme constaté lors de nos études, la malnutrition aigüe est largement répandue parmi les enfants d’âge scolaire et les adolescents. Les dernières directives de l’OMS pour le traitement de la malnutrition aigüe sévère recommandent d’établir des points de coupure pour le périmètre brachial (PB) chez les enfants de plus de 5 ans. En effet, bien que la malnutrition soit moins liée à la mortalité que chez les jeunes enfants, lutter contre la malnutrition à l’âge scolaire est important car elle entrave le développement physique et mental également dans cette tranche d’âge.

· Points de coupure du périmètre brachial (PB) pour détecter la malnutrition aigue chez les enfants d’age scolaire et les adolescents cambodgiens

Les données anthropométriques de la base de données du Cambodge furent utilisées. Des courbes ROC (« Receiver operating characteristic ») furent établies pour comparer la performance des points de coupure du PB au« golden standard » recommandé par l’OMS pour définir la malnutrition aigüe sévère et modérée (BAZ <-2 et BAZ <-3). Les points de coupure avec la meilleure sensibilité et un taux de faux positifs inférieur à 33% furent sélectionnés comme les points de coupure de PB optimaux pour dépister la malnutrition aigüe parmi les points de coupure avec les valeurs d’aire sous la courbe (AUC) les plus élevées.

18.4 18.2 18 17.8 17.6 17.4 17.2 17 16.8 16.6 16.4 16.2 16 15.8 15.6 15.4

PB (cm) PB 15.2 15 14.8 14.6 14.4 14.2 14 13.8 13.6 13.4 13.2 13 0 - 1.9 2 - 4.9 5 - 7.9 8 - 10.9 11 - 13.9

Age (années) MA garçons MA filles MA tous MAS garçons MAS filles MAS tous

Points de coupure du PB optimaux pour détecter la malnutrition aigüe et la malnutrition aigüe sévère par tranche d’âge et p ar sexe

Les points de coupure optimaux augmentèrent graduellement avec l’âge de façon quasi -linéaire, mais cette évolution est influencée par les fluctuations de la vitesse de croissance, notamment pendant la puberté. Les points de coupure furent plus élevés chez les garçons que chez les filles (+0.1 +0.3 cm), sauf dans la tranche d’âge 8 -10.9 ans. Nous supposons que le pic de croissance plus précoce chez les filles que chez les garçons et associés à des modifications des masses maigre et grasse explique ce phénomène.

170

La prise en charge de la malnutrition aigüe devrait prendre en compte les enfants âgés de plus de 5 ans. En effet, à l’âge scolaire, la malnutrition aigüe, qui va souvent de pair avec des carences en micronutriments, peut retarder la maturation sexuelle et altérer la puissance musculaire, la densité osseuse et la capacité de travail. La malnutrition à l’âge scolaire augmente la morbidité, le risque d’échec et d’abandon scolaire. Les écoles peuvent être une plateforme efficace pour le suivi de la malnu trition aigüe chez les enfants d’âge scolaire.

· Influence de l’inflammation subclinique sur les concentrations plasmatiques en récepteurs solubles de la transferrine, ferritine, RBP et zinc chez les enfants d’âge scolaire au Cambodge et au Sénégal

La phase inflammatoire est connue pour altérer les concentrations plasmatiques en ferritine et en rétinol alors que le statut en micronutriment reste inchangé, conduisant à respectivement sous-estimer et surestimer les carences en fer et en vitamine A. La TFR a longtemps été considérée comme non sensible ou moins sensible à l’inflammation que la ferritine mais des recherches récentes ont mis en évidence a impact significatif de la phase inflammatoire sur les concentrations en TFR. L’effet de l’inflammation sur le s concentrations plasmatiques en zinc demeure également incertain. Nous avons utilisé les données du Cambodge et du Sénégal pour évaluer l’impact de la phase inflammatoire sur l’évaluation des carences en micronutriments chez les enfants d’âge scolaire. Le ratio des valeurs moyennes du biomarqueur chez les groupes en différentes phases inflammatoires par rapport au groupe sans inflammation fut calculé. Le facteur de correction égal à 1/ ratio fut appliqué aux valeurs du biomarqueurs pour déterminer la prévalence des carences en micronutriments ajustés pour l’inflammation. Les prévalences furent aussi ajustées avec les facteurs de correction pour la ferritine et le RBP proposés par Thurnham et al.

Ratios et facteurs de correction (FC) en fonction de l’infl ammation pour la ferritine, les récepteurs de la transferrine, le RBP et le zinc Ferritine Recepteur soluble transferrine Ratio (95%) 1 FC 2 Ratio (95%) 1 FC 2 Enfants sénégalais Incubation 1.22 [1.16-1.26] 0.82 0.97 [0.32-1.63] 1.03 Convalescence 2.00 [1.96-2.03] 0.50 1.07 [1.00-1.13] 0.94 Convalescence tardive 1.50 [1.47-1.52] 0.67 1.02 [0.97-1.08] 0.98 Enfants cambodgiens Incubation 1.29 [1.27-1.31] 0.77 0.90 [0.79-1.01] 1.12 Convalescence 1.83 [1.82-1.83] 0.55 1.10 [1.07-1.13] 0.91 Convalescence tardive 1.27 [1.27-1.27] 0.79 1.15 [1.15-1.18] 0.87 RBP Zinc Ratio (95%) 1 FC 2 Ratio (95%) 1 FC 2 Enfants sénégalais Incubation 0.76 [0.64-0.88] 1.31 1.05 [0.96-1.08] 0.95 Convalescence 0.86 [0.75-0.97] 1.16 0.96 [0.95-1.02] 1.04 Convalescence tardive 0.94 [0.85-1.03] 1.07 1.01 [0.98-1.03] 0.99 Enfants cambodgiens Incubation 0.85 [0.63-1.07] 1.17 0.93 [0.88-1.05] 1.07 Convalescence 0.85 [0.82-0.89] 1.17 1.00 [0.98-1.02] 1.00 Convalescence tardive 1.05 [1.03-1.08] 0.95 1.01 [1.00-1.02] 0.99 1 Ratio des concentrations dans le groupe en inflammation vs. le groupe sans inflammation 2 FC =1/Ratio

171

Enfants sénégalais Enfants cambodgiens

CarenceZinc en deficiency zinc Zinc deficiency

Statut marginalMarginal VAen statusVA * Marginal VA *status *

Vitamin A deficiency Carence en VA VAD * ID (high index) Carence en fer (index élevé) ID (high index) *

* ID (low body iron) * ID (low body iron) * Carence en fer (faible fer * corporel) * * ID (highTfR/lowFER) ID (highTfR/lowFER) * Carence en fer (TFR élevée/ * * FER basse)Low FER * Low FER * * *

FER basse High TfR High TfR *

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 0 5 10 15 20 25 30 35 40 45 Prevalence (%) TFR élevée Prevalence (%) Uncorrected

Point de coupure ferritine 30 µg/L pour les enfants avec inflammation Corrigé avec les FC de Thurnham Corrigé avec nos FC (3 phases d'inflammation) Enfants sans inflammation Non corrigé * P <0.05 Test de Mc Nemar pour comparer prevalence ajustée vs prévalence non ajustée VA: Vitamin A

Influence de la correction de TFR, FER, RBP, et zinc en fonction de l’inflammation sur la prévalence des carences en micronutriments

Dans des recherches précédentes, TFR était considéré comme plus fiable que la ferritine dans les zones ou l’inflammation est endémique, mais nous avons montré que TFR est également sensible à la phase inflammatoire et qu’elle doit être ajustée en fonction de l’inflammation. Nos analyses ont montré que l’inflammation augmente la ferritine et les récepteurs de la transferrin, conduisant à sous-estimer la prevalence des stocks de fer insuffisants (taux de ferritine bas) et à surestimer la prevalence de la carence des tissus en fer (taux de TFR élevés), faisant de l’index TFR/log FER l’indicateur le moins sensible à l’inflammation. Le RBP diminue pendant l’inflammation ainsi les prévalences de carence en vitamine A et de statut marginal en v itamine A sont surestimées dans les populations avec des taux d’inflammation élevés. Nos résultats suggèrent que pour déterminer le statut en zinc des populations ou l’inflammation est sub - clinique, prendre en compte la phase inflammatoire ne modifiera pas significativement l’estimation de la prévalence de la carence en zinc. Globalement, plus de recherche est nécessaire à propose des indicateurs de statut nutritionnel des enfants d’âge scolaire et es adolescents. La variabilité intra-populations, les effets des facteurs environnementaux et les besoins spécifiques de cette tranche d’âge devraient être pris en compte dans la mise au point des indicateurs.

172

Conclusion

Des associations positives entre carences en micronutriments, infection parasitaire and faibles performance cognitives furent observées au Cambodge, mettant en évidence la nécessité de traiter les problèmes de nutrition et de santé chez les enfants d’âge scolaire et les adolescents. Même si la recherche cible rarement les enfants d’âge sco laire et les adolescents, on sait que les besoins nutritionnels élevés associés au pic de croissance et de développement cérébral en font une population particulièrement vulnérable aux carences nutritionnelles, ce qui compromet la réussite scolaire, pourtant cruciale pour rompre le cycle intergénérationnel de la malnutrition et de la pauvreté. Le statut en vitamine A et les performances cognitives ont été améliorés grâce au riz fortifié dans notre étude au Cambodge, montrant que certains des effets négatifs de la malnutrition peuvent encore être corrigés à l’âge scolaire. De faibles apports nutritionnels et des taux élevés de malnutrition aigüe et de carences en micronutriments chez des enfants urbains ne bénéficiant pas d’alimentation scolaire comme nous l ’avons observé au Sénégal mettent en évidence le besoin d’interventions à l’école. L’éducation à la nutrition ainsi que l’alimentation scolaire devraient être mis en place dans un nombre croissant d’écoles, notamment dans les zones urbaines défavorisées. Pour convaincre les décideurs politiques de la nécessité d’étendre ces programmes d’alimentation scolaire, de solides preuves scientifiques de leurs bénéfices sur la santé et la nutrition sont nécessaires. Des effets bénéfiques prometteurs sur les statuts en micronutriments et les performances cognitives furent observés après seulement 6 mois de consommation de riz fortifié lors de notre recherche. Un impact plus important peut être attendu de la consommation quotidienne à plus long terme de riz fortifié. A l’avenir, des études d’efficacités conduites sur plus d’une année scolaire pourrait favoriser l’introduction d’aliments de base fortifiés dans les programmes existants d’alimentation scolaire. En outre, une recherche plus approfondie prenant en compte la variabilité intra-populations, l’influence des facteurs environnementaux et les besoins spécifiques des enfants d’âge scolaire et des adolescents est nécessaire pour mieux identifier et caractériser la malnutrition dans cette tranche d’âge. Nous avons établ i des points de coupure pour le périmètre brachial au Cambodge, un outil peu couteux pour détecter la malnutrition aigüe chez les enfants scolaires. Nous avons proposé des facteurs de correction pour éliminer l’effet de confusion de la phase inflammatoire sur les concentrations plasmatiques de récepteurs de la transferrine, et vérifié les facteurs de correction des statuts en ferritine et en vitamine A. Négliger le problème de santé publique que constitue la malnutrition chez les enfants d’âge scolaire et les adolescents risque de compromettre le progrès historiques accompli chez les jeunes enfants. Investir dans la nutrition de ces populations est une opportunité d’optimiser le développement des générations futures dans les pays en développement.

173

References

1. WHO. Global nutrition policy review: what does it take to scale up nutrition action? : World Health Organization; 2013. 2. Berger J, Greffeuille V, Dijkhuizen M, Wieringa F. Aliments d’origine animale et nutrition des populations vulnérables des pays du Sud. Bulletin de l'Académie vétérinaire de France. 2013;166(4):319-327. 3. Black RE, Victora CG, Walker SP, et al. Maternal and child undernutrition and overweight in low-income and middle-income countries. The Lancet. 2013;382(9890):427-451. 4. Ahmed T, Hossain M, Sanin KI. Global burden of maternal and child undernutrition and micronutrient deficiencies. Annals of Nutrition and Metabolism. 2012;61(Suppl. 1):8-17. 5. Zimmermann MB, Jooste PL, Pandav CS. Iodine-deficiency disorders. Lancet. 2008;372(9645):1251-1262. 6. Haas JD, Brownlie T. Iron deficiency and reduced work capacity: a critical review of the research to determine a causal relationship. The Journal of nutrition. 2001;131(2):676S- 690S. 7. Walker CF, Ezzati M, Black R. Global and regional child mortality and burden of disease attributable to zinc deficiency. European journal of clinical nutrition. 2009;63(5):591- 597. 8. Whitcher JP, Srinivasan M, Upadhyay MP. Corneal blindness: a global perspective. Bulletin of the World Health Organization. 2001;79(3):214-221. 9. Sommer A, Katz J, Tarwotjo I. Increased risk of respiratory disease and diarrhea in children with preexisting mild vitamin A deficiency. The American journal of clinical nutrition. 1984;40(5):1090-1095. 10. Bailey RL, West Jr KP, Black RE. The epidemiology of global micronutrient deficiencies. Annals of Nutrition and Metabolism. 2015;66(Suppl. 2):22-33. 11. Ahmed AU. Impact of feeding children in school: Evidence from Bangladesh. Washington, DC: International Food Policy Research Institute. 2004. 12. Bhutta ZA, Das JK, Rizvi A, et al. Evidence-based interventions for improvement of maternal and child nutrition: what can be done and at what cost? The Lancet. 2013;382(9890):452-477. 13. Engle PL, Black MM, Behrman JR, et al. Strategies to avoid the loss of developmental potential in more than 200 million children in the developing world. The lancet. 2007;369(9557):229-242. 14. WHO. The second decade. Improving Adolescent Health and Development. 2001. 15. Kurz KM. Adolescent nutritional status in developing countries. Proceedings of the Nutrition Society. 1996;55(1B):319-331. 16. Organization WH. Adolescent nutrition: a review of the situation in selected south-east Asian countries. SEA-NUT-163. New Delhi: WHO, Regional Office for South East Asia. 2006. 17. Best C, Neufingerl N, van Geel L, van den Briel T, Osendarp S. The nutritional status of school-aged children: why should we care? Food Nutr Bull. 2010;31(3):400-417. 18. WHO. Physical status: the use and interpretation of anthropometry : report of a WHO Expert Committee. WHO Technical Report Series : World Health Organization; 1995. 19. Shetty PS, James W. Body mass index. A measure of chronic energy deficiency in adults. FAO food and nutrition paper. 1993;56:1-57. 20. Neumann CG, Murphy SP, Gewa C, Grillenberger M, Bwibo NO. Meat supplementation improves growth, cognitive, and behavioral outcomes in Kenyan children. the Journal of Nutrition. 2007;137(4):1119-1123. 21. Muthayya S, Eilander A, Transler C, et al. Effect of fortification with multiple micronut rients and n− 3 fatty acids on growth and cognitive performance in Indian schoolchildren: the CHAMPION (Children's Health and Mental Performance Influenced

174

by Optimal Nutrition) Study. The American journal of clinical nutrition. 2009;89(6):1766- 1775. 22. Sachdev H, Gera T, Nestel P. Effect of iron supplementation on mental and motor development in children: systematic review of randomised controlled trials. Public Health Nutr. 2005;8(02):117-132. 23. Zimmermann MB, Connolly K, Bozo M, Bridson J, Rohner F, Grimci L. Iodine supplementation improves cognition in iodine-deficient schoolchildren in Albania: a randomized, controlled, double-blind study. Am J Clin Nutr. 2006;83(1):108-114. 24. Stoltzfus RJ, Albonico M, Tielsch JM, Chwaya HM, Savioli L. Linear growth retardation in Zanzibari school children. The Journal of nutrition. 1997;127(6):1099-1105. 25. Jukes MC, Drake LJ, Bundy DA. School health, nutrition and education for all: levelling the playing field. CABI; 2007. 26. Unicef. The State of the World's Children 2011-Executive Summary: Adolescence an Age of Opportunity. Unicef; 2011. 27. International Wv. Before she's ready. 15 places girls marry by 15. 2008. 28. United Nations DoEaSA, Population Division. World Population Prospects: the 2015 revision. 2015; < www.esa.un.org/unpd/wpp2008/index.htm> . Accessed August 2015. 29. Programme WF. State of School Feeding Worldwide 2013 Overview. 2013. 30. Wells JC. Toward body composition reference data for infants, children, and adolescents. Advances in Nutrition: An International Review Journal. 2014;5(3):320S-329S. 31. Soliman A, De Sanctis V, Elalaily R. Nutrition and pubertal development. Indian journal of endocrinology and metabolism. 2014;18(Suppl 1):S39. 32. Story M, Stang J. Nutrition needs of adolescents. Guidelines for adolescent nutrition services. Minneapolis, MN: Centre for Leadership, Education and Training in Maternal and Child Nutrition Division of Epidemiology and Community Health, School of Public Health, University of Minnesota. 2005:21-34. 33. Hallberg L, Rossander-Hulten L. Iron requirements in menstruating women. The American journal of clinical nutrition. 1991;54(6):1047-1058. 34. Thompson RA, Nelson CA. Developmental science and the media: Early brain development. American Psychologist. 2001;56(1):5. 35. Benton D. The influence of dietary status on the cognitive performance of children. Molecular nutrition & food research. 2010;54(4):457-470. 36. Bryan J, Osendarp S, Hughes D, Calvaresi E, Baghurst K, van Klinken JW. Nutrients for cognitive development in school-aged children. Nutr Rev. 2004;62(8):295-306. 37. Isaacs E, Oates J. Nutrition and cognition: assessing cognitive abilities in children and young people. Eur J Nutr. 2008;47 Suppl 3:4-24. 38. Giedd JN, Blumenthal J, Jeffries NO, et al. Brain development during childhood and adolescence: a longitudinal MRI study. Nature neuroscience. 1999;2(10):861-863. 39. Hoffman DJ. Growth retardation and metabolic programming: implications and consequences for adult health and disease risk. Jornal de pediatria. 2014;90(4):325-328. 40. Proos LA, Hofvander Y, Tuvemo T. Menarcheal Age and Growth Pattern of Indian Girls Adopted in Sweden: I. Menarcheal Age. Acta Paediatrica. 1991;80(8-9):852 -858. 41. Cook S, Weitzman M, Auinger P, Nguyen M, Dietz WH. Prevalence of a metabolic syndrome phenotype in adolescents: findings from the third National Health and Nutrition Examination Survey, 1988-1994. Archives of pediatrics & adolescent medicine. 2003;157(8):821-827. 42. Association AD. Type 2 diabetes in children and adolescents. Pediatrics. 2000;105(3):671-680. 43. Vanhala M, Vanhala P, Kumpusalo E, Halonen P, Takala J. Relation between obesity from childhood to adulthood and the metabolic syndrome: population based study. Bmj. 1998;317(7154):319-320. 44. Latham MC. Human nutrition in the developing world. Food & Agriculture Org.; 1997.

175

45. Nokes C, van den Bosch C, Bundy DA. The effects of iron deficiency and anemia on mental and motor performance, educational achievement, and behavior in children. A report of the INACG. Washington, DC: International Life Sciences Institute. 1998. 46. Pasricha S-R, Drakesmith H, Black J, Hipgrave D, Biggs B-A. Control of iron deficiency anemia in low-and middle-income countries. Blood. 2013;121(14):2607-2617. 47. Grantham-McGregor S, Ani C. A review of studies on the effect of iron deficiency on cognitive development in children. J Nutr. 2001;131(2):649S-668S. 48. Sungthong R, Mo-suwan L, Chongsuvivatwong V. Effects of hae moglobin and serum ferritin on cognitive function in school children. Asia Pac J Clin Nutr. 2002;11(2):117- 122. 49. Halterman JS, Kaczorowski JM, Aligne CA, Auinger P, Szilagyi PG. Iron deficiency and cognitive achievement among school-aged children and adolescents in the United States. Pediatrics. 2001;107(6):1381-1386. 50. Otero GA, Pliego-Rivero FB, Porcayo-Mercado R, Mendieta-Alcántara G. Working memory impairment and recovery in iron deficient children. Clinical Neurophysiology. 2008;119(8):1739-1746. 51. McCann JC, Ames BN. An overview of evidence for a causal relation between iron deficiency during development and deficits in cognitive or behavioral function. Am J Clin Nutr. 2007;85(4):931-945. 52. Chen M-H, Su T-P, Chen Y-S, et al. Association between psychiatric disorders and iron deficiency anemia among children and adolescents: a nationwide population-based study. BMC psychiatry. 2013;13(1):161. 53. Bruner AB, Joffe A, Duggan AK, Casella JF, Brandt J. Randomised study of cognitive effects of iron supplementation in non-anaemic iron-deficient adolescent girls. Lancet. 1996;348(9033):992-996. 54. Lynn R, Harland EP. A positive effect of iron supplementation on the IQs of iron deficient children. Personality and Individual Differences. 1998;24(6):883-885. 55. Hermoso M, Vucic V, Vollhardt C, et al. The effect of iron on cognitive development and function in infants, children and adolescents: a systematic review. Ann Nutr Metab. 2011;59(2-4):154-165. 56. Low M, Farrell A, Biggs B-A, Pasricha S-R. Effects of daily iron supplementation in primary-school –aged children: systematic review and meta-analysis of randomized controlled trials. Canadian Medical Association Journal. 2013:cmaj. 130628. 57. Palti H, Meijer A, Adler B. Learning achievement and behavior at school of anemic and non-anemic infants. Early human development. 1985;10(3):217-223. 58. Scrimshaw NS. Functional consequences of iron deficiency in human populations. Journal of Nutritional Science and Vitaminology. 1984;30(1):47-63. 59. Ayoya MA, Spiekermann-Brouwer GM, Traoré AK, Garza C. Effect on school attendance and performance of iron and multiple micronutrients as adjunct to drug treatment of Schistosoma-infected anemic schoolchildren. Food & Nutrition Bulletin. 2012;33(4):235- 241. 60. Murray-Kolb LE, Beard JL. Iron treatment normalizes cognitive functioning in young women. Am J Clin Nutr. 2007;85(3):778-787. 61. Georgieff MK. Nutrition and the developing brain: nutrient priorities and measurement. The American journal of clinical nutrition. 2007;85(2):614S-620S. 62. Bleichrodt N, Born MP. A meta-analysis of research on iodine and its relationship to cognitive development. The damaged brain of iodine deficiency, New York: Cognizant Communication. 1994:195-200. 63. Qian M, Wang D, Watkins WE, et al. The effects of iodine on intelligence in children: a meta-analysis of studies conducted in China. Asia Pacific journal of clinical nutrition. 2005;14(1):32-42. 64. van den Briel T, West CE, Bleichrodt N, van de Vijver FJ, Ategbo EA, Hautvast JG. Improved iodine status is associated with improved mental performance of schoolchildren in Benin. Am J Clin Nutr. 2000;72(5):1179-1185.

176

65. Pardede LV, Hardjowasito W, Gross R, et al. Urinary iodine excretion is the most appropriate outcome indicator for iodine deficiency at field conditions at district level. The Journal of nutrition. 1998;128(7):1122-1126. 66. Dodge PR, Palkes H, Fierro-Benitez R, Ramirez I. Effect on intelligence of iodine in oil administered to young Andean children-a preliminary report. Endemic goiter. Washington, DC: Pan American Health Organization. 1969:378-380. 67. Bautista A, Barker PA, Dunn JT, Sanchez M, Kaiser DL. The effects of oral iodized oil on intelligence, thyroid status, and somatic growth in school-age children from an area of endemic goiter. Am J Clin Nutr. 1982;35(1):127-134. 68. Shrestha R. Effect of iodine and iron supplementation on physical, psychomotor and mental development in primary school children in Malawi. Landbouwuniversiteit te Wageningen; 1994. 69. Semba RD, Piot P, Bloem MW. Nutrition and health in developing countries. Springer; 2008:377-434. 70. Thornton KA, Mora-Plazas M, Marín C, Villamor E. Vitamin A deficiency is associated with gastrointestinal and respiratory morbidity in school-age children. The Journal of nutrition. 2014;144(4):496-503. 71. Bloem MW, Wedel M, EGGER RJ, et al. Mild vitamin A deficiency and risk of respiratory tract diseases and diarrhea in preschool and school children in northeastern Thailand. American journal of epidemiology. 1990;131(2):332-339. 72. Brabin L, Brabin BJ. The cost of successful adolescent growth and development in girls in relation to iron and vitamin A status. The American journal of clinical nutrition. 1992;55(5):955-958. 73. Leenstra T, Kariuki S, Kurtis J, Oloo A, Kager P, Ter Kuile F. The effect of weekly iron and vitamin A supplementation on hemoglobin levels and iron status in adolescent schoolgirls in western Kenya. European journal of clinical nutrition. 2009;63(2):173-182. 74. Mwanri L, Worsley A, Ryan P, Masika J. Supplemental vitamin A improves anemia and growth in anemic school children in Tanzania. The Journal of nutrition. 2000;130(11):2691-2696. 75. King JC. Zinc: an essential but elusive nutrient. The American journal of clinical nutrition. 2011;94(2):679S-684S. 76. Gibson R, Vanderkooy P, MacDonald A, Goldman A, Ryan B, Berry M. A growth-limiting, mild zinc-deficiency syndrome in some southern Ontario boys with low height percentiles. The American journal of clinical nutrition. 1989;49(6):1266-1273. 77. Castillo-Duran C, García H, Venegas P, et al. Zinc supplementation increases growth velocity of male children and adolescents with short stature. Acta Paediatrica. 1994;83(8):833-837. 78. Porto S, Oliveira HP, Cunha AJ, et al. Linear growth and zinc supplementation in children with short stature. Journal of Pediatric Endocrinology and Metabolism. 2000;13(8):1121- 1128. 79. Baqui AH, Black RE, El Arifeen S, et al. Effect of zinc supplementation started during diarrhoea on morbidity and mortality in Bangladeshi children: community randomised trial. Bmj. 2002;325(7372):1059. 80. Mayo-Wilson E, Junior JA, Imdad A, et al. Zinc supplementation for preventing mortality, morbidity, and growth failure in children aged 6 months to 12 years of age. The Cochrane Library. 2014. 81. Bhutta ZA, Bird SM, Black RE, et al. Therapeutic effects of oral zinc in acute and persistent diarrhea in children in developing countries: pooled analysis of randomized controlled trials. The American journal of clinical nutrition. 2000;72(6):1516-1522. 82. Kawade R. Zinc status and its association with the health of adolescents: a review of studies in India. Global health action. 2012;5. 83. Butterworth RF. Thiamin deficiency and brain disorders. Nutrition research reviews. 2003;16(02):277-284.

177

84. Kornreich L, Bron-Harlev E, Hoffmann C, et al. Thiamine deficiency in infants: MR findings in the brain. American journal of neuroradiology. 2005;26(7):1668-1674. 85. Joint F, Organization WH. Vitamin and mineral requirements in human nutrition. 2005. 86. Fattal I, Friedmann N, Fattal-Valevski A. The crucial role of thiamine in the development of syntax and lexical retrieval: a study of infantile thiamine deficiency. Brain. 2011;134(6):1720-1739. 87. Herbison CE, Hickling S, Allen KL, et al. Low intake of B-vitamins is associated with poor adolescent mental health and behaviour. Preventive medicine. 2012;55(6):634-638. 88. Lee BY, Yanamandra K, Bocchini JA. Thiamin deficiency: A possible major cause of some tumors?(Review). Oncology reports. 2005;14(6):1589-1592. 89. Mistry HD, Williams PJ. The importance of antioxidant micronutrients in pregnancy. Oxidative medicine and cellular longevity. 2011;2011. 90. Knovich MA, Il'yasova D, Ivanova A, Molnár I. The association between serum copper and anaemia in the adult Second National Health and Nutrition Examination Survey (NHANES II) population. British journal of nutrition. 2008;99(06):1226-1229. 91. Harvey LJ, Ashton K, Hooper L, Casgrain A, Fairweather-Tait SJ. Methods of assessment of copper status in humans: a systematic review. The American journal of clinical nutrition. 2009:ajcn. 27230E. 92. Becker DA, Balcer LJ, Galetta SL. The neurological complications of nutritional deficiency following bariatric surgery. Journal of obesity. 2012;2012. 93. Cordano A. Clinical manifestations of nutritional copper deficiency in infants and children. The American journal of clinical nutrition. 1998;67(5):1012S-1016S. 94. Dror DK, Allen LH. Vitamin E deficiency in developing countries. Food & Nutrition Bulletin. 2011;32(2):124-143. 95. Muller D, Lloyd J, Wolff O. Vitamin E and neurological function. The Lancet. 1983;321(8318):225-228. 96. Koscik RL, Lai HJ, Laxova A, et al. Preventing early, prolonged vitamin E deficiency: an opportunity for better cognitive outcomes via early diagnosis through neonatal screening. The Journal of pediatrics. 2005;147(3):S51-S56. 97. Fred Gey K. Vitamins E plus C and interacting conutrients required for optimal health. BioFactors. 1998;7(1-2):113 -174. 98. Tarwotjo I, Katz J, West K, Tielsch JM, Sommer A. Xerophthalmia and growth in preschool Indonesian children. The American journal of clinical nutrition. 1992;55(6):1142-1146. 99. Singh S. Adolescent childbearing in developing countries: a global review. Studies in family planning. 1998:117-136. 100. Thurnham DI. Nutrition of adolescent girls in low-and middle-income countries. Sight Life. 2013;27:26-37. 101. Organization WH. Adolescent pregnancy. http://who.int/mediacentre/factsheets/fs364/en/ . Available at. Accessed 08/05/2015. 102. Kennedy E, Gray N, Azzopardi P, Creati M. Adolescent fertility and family planning in East Asia and the Pacific: a review of DHS reports. Reproductive health. 2011;8(1):11. 103. Haider BA, Olofin I, Wang M, Spiegelman D, Ezzati M, Fawzi WW. Anaemia, prenatal iron use, and risk of adverse pregnancy outcomes: systematic review and meta-analysis. Bmj. 2013;346. 104. Dijkhuizen MA, Wieringa FT, West CE, Muherdiyantiningsih, Muhilal. Concurrent micronutrient deficiencies in lactating mothers and their infants in Indonesia. Am J Clin Nutr. 2001;73:786-791. 105. Christian P, Stewart CP. Maternal micronutrient deficiency, fetal development, and the risk of chronic disease. The Journal of nutrition. 2010;140(3):437-445. 106. Kurth F, Bélard S, Mombo-Ngoma G, et al. Adolescence as risk factor for adverse pregnancy outcome in Central Africa-a cross-sectional study. PLoS One. 2010;5(12):e14367.

178

107. Wallace JM, Milne JS, Aitken RP. Effect of weight and adiposity at conception and wide variations in gestational dietary intake on pregnancy outcome and early postnatal performance in young adolescent sheep. Biology of reproduction. 2010;82(2):320-330. 108. Williamson C. Nutrition in pregnancy. Nutrition bulletin. 2006;31(1):28-59. 109. Baker PN, Wheeler SJ, Sanders TA, et al. A prospective study of micronutrient status in adolescent pregnancy. The American journal of clinical nutrition. 2009;89(4):1114-1124. 110. Young BE, McNanley TJ, Cooper EM, et al. Maternal vitamin D status and calcium intake interact to affect fetal skeletal growth in utero in pregnant adolescents. The American journal of clinical nutrition. 2012;95(5):1103-1112. 111. Rosso P. Nutrition and maternal-fetal exchange. American journal of clinical nutrition. 1981. 112. Roche AF, Davila GH. Late adolescent growth in stature. Pediatrics. 1972;50(6):874-880. 113. Spear BA. Adolescent growth and development. Journal of the American Dietetic Association. 2002;102(3):S23-S29. 114. Riley AP, Huffman SL, Chowdhury A. Age at menarche and postmenarcheal growth in rural Bangladeshi females. Annals of human biology. 1989;16(4):347-359. 115. Baumgartner RN, Roche AF, Himes JH. Incremental growth tables: supplementary to previously published charts. The American journal of clinical nutrition. 1986;43(5):711- 722. 116. Moerman ML. Growth of the birth canal in adolescent girls. Obstetrics and Gynecology. 1982;143:528-532. 117. King JC. The risk of maternal nutritional depletion and poor outcomes increases in early or closely spaced pregnancies. The Journal of nutrition. 2003;133(5):1732S-1736S. 118. Rah JH, Christian P, Shamim AA, Arju UT, Labrique AB, Rashid M. Pregnancy and lactation hinder growth and nutritional status of adolescent girls in rural Bangladesh. The Journal of nutrition. 2008;138(8):1505-1511. 119. Wallace J, Bourke D, Da Silva P, Aitken R. Nutrient partitioning during adolescent pregnancy. Reproduction. 2001;122(3):347-357. 120. Taggart NR, Holliday RM, Billewicz W, Hytten FE, Thomson AM. Changes in skinfolds during pregnancy. British journal of nutrition. 1967;21(02):439-451. 121. Scholl TO, Hediger ML, Schall JI, Khoo C-S, Fischer RL. Maternal growth during pregnancy and the competition for nutrients. The American journal of clinical nutrition. 1994;60(2):183-188. 122. Thame MM, Fletcher HM, Baker TM, Jahoor F. Comparing the glucose kinetics of adolescent girls and adult women during pregnancy. The American journal of clinical nutrition. 2010;91(3):604-609. 123. Wallace J, Luther J, Milne J, et al. Nutritional modulation of adolescent pregnancy outcome –a review. Placenta. 2006;27:61-68. 124. Organization/UNICEF/ICCIDD WH. Assessment of Iodine Deficiency Disorders and Monitoring their Elimination. A guide for program managers. 3rd ed. Geneva: WHO; 2007. 125. Organization WH. Serum retinol concentrations for determining the prevalence of vitamin A deficiency in populations. 2011. 126. de Benoist B, Darnton-Hill I, Davidsson L, Fontaine O, Hotz C. Conclusions of the Joint WHO/UNICEF/IAEA/IZiNCG Interagency Meeting on Zinc. Food Nutr Bull. 2007;28(3 Suppl):S480-484. 127. Organization WH. Serum ferritin concentrations for the assessment of iron status and iron deficiency in populations. 2011. 128. WHO. Interpretation Guide. Nutrition Landscape Information System. Country profile indicators. Geneva: World Health Organization; 2010. 129. Organization WH. Serum and red blood cell folate concentrations for assessing folate status in populations. 2015.

179

130. Thankachan P, Rah JH, Thomas T, et al. Multiple micronutrient-fortified rice affects physical performance and plasma vitamin B-12 and homocysteine concentrations of Indian school children. The Journal of nutrition. 2012;142(5):846-852. 131. Shaw N, Wang J, Pan W, Liao P, Yang FL. Thiamin and riboflavin status of Taiwanese elementary schoolchildren. Asia Pacific journal of clinical nutrition. 2007;16:564. 132. Hotez PJ, Bethony JM, Diemert DJ, Pearson M, Loukas A. Developing vaccines to combat hookworm infection and intestinal schistosomiasis. Nature Reviews Microbiology. 2010;8(11):814-826. 133. Khounnorath S, Chamberlain K, Taylor AM, et al. Clinically unapparent infantile thiamin deficiency in Vientiane, Laos. 2011. 134. Harper C. Thiamine (vitamin B1) deficiency and associated brain damage is still common throughout the world and prevention is simple and safe! European Journal of Neurology. 2006;13(10):1078-1082. 135. Smith LC, Ruel MT, Ndiaye A. Why is child malnutrition lower in urban than in rural areas? Evidence from 36 developing countries. World Development. 2005;33(8):1285- 1305. 136. Fotso J-C. Urban –rural differentials in child malnutrition: trends and socioeconomic correlates in sub-Saharan Africa. Health & Place. 2007;13(1):205-223. 137. Menon P, Ruel MT, Morris SS. Socio-economic differentials in child stunting are consistently larger in urban than rural areas: analysis of 10 DHS data sets. FASEB J. 1999;13(4):A543. 138. Ruel MT, Garrett JL, Hawkes C, Cohen MJ. The food, fuel, and financial crises affect the urban and rural poor disproportionately: a review of the evidence. the Journal of Nutrition. 2010;140(1):170S-176S. 139. Tinker I. Street foods: Urban food and employment in developing countries. Oxford University Press on Demand; 1997. 140. Zezza A, Tasciotti L. Urban agriculture, poverty, and food security: Empirical evidence from a sample of developing countries. Food policy. 2010;35(4):265-273. 141. Popkin BM. Global nutrition dynamics: the world is shifting rapidly toward a diet linked with noncommunicable diseases. The American journal of clinical nutrition. 2006;84(2):289-298. 142. Brinkman H-J, de Pee S, Sanogo I, Subran L, Bloem MW. High food prices and the global financial crisis have reduced access to nutritious food and worsened nutritional status and health. The Journal of nutrition. 2010;140(1):153S-161S. 143. Torlesse H, Kiess L, Bloem MW. Association of household rice expenditure with child nutritional status indicates a role for macroeconomic food policy in combating malnutrition. The Journal of Nutrition. 2003;133(5):1320-1325. 144. Jensen RT, Miller NH. The impact of the world food price crisis on nutrition in China. 2008. 145. Benkhala A, Bastard G, Broutin C. Rechercher des réponses viables aux défis de la nutrition des populations vulnérables, synthèse de l'étude exploratoire,. Danone Communities, GRET, IFAN, Enda-Graf; 2009. 146. Sulaiman M, Parveen M, Das NC. Impact of the food price hike on nutritional status of women and children. Research and Evaluation Division, BRAC; 2009. 147. West KP, Mehra S. Vitamin A intake and status in populations facing economic stress. The Journal of nutrition. 2010;140(1):201S-207S. 148. FAO. The state of food insecurity in the world. High food prices and food security - threats and opportunities. Rome2008. 149. Omemu A, Aderoju S. Food safety knowledge and practices of street food vendors in the city of Abeokuta, Nigeria. Food control. 2008;19(4):396-402. 150. Muyanja C, Nayiga L, Brenda N, Nasinyama G. Practices, knowledge and risk factors of street food vendors in Uganda. Food Control. 2011;22(10):1551-1558. 151. Ochola S, Masibo PK. Dietary intake of schoolchildren and adolescents in developing countries. Ann Nutr Metab. 2014;64 Suppl 2:24-40.

180

152. Shi Z, Lien N, Kumar BN, Holmboe-Ottesen G. Socio-demographic differences in food habits and preferences of school adolescents in Jiangsu Province, China. European Journal of Clinical Nutrition. 2005;59(12):1439-1448. 153. Pfäffle R. Genetics of growth in the normal child. European journal of endocrinology. 2006;155(suppl 1):S27-S33. 154. Butte NF, Garza C, de Onis M. Evaluation of the feasibility of international growth standards for school-aged children and adolescents. The Journal of nutrition. 2007;137(1):153-157. 155. Habicht J-P, Yarbrough C, Martorell R, Malina R, Klein R. Height and weight standards for preschool children: How relevant are ethnic differences in growth potential? The Lancet. 1974;303(7858):611-615. 156. Haas JD, Campirano F. Interpopulation variation in height among children 7 to 18 years of age. Food & Nutrition Bulletin. 2006;27(Supplement 5):212-223. 157. de Onis M, Onyango AW, Borghi E, Siyam A, Nishida C, Siekmann J. Development of a WHO growth reference for school-aged children and adolescents. Bull World Health Organ. 2007;85(9):660-667. 158. Frisancho AR. New norms of upper limb fat and muscle areas for assessment of nutritional status. 159. Cogill B, Food, Project NTA. Anthropometric Indicators Measurement Guide. Food and Nutrition Technical Assistance Project, Academy for Educational Development; 2003. 160. Ozturk A, Budak N, Cicek B, Mazicioglu MM, Bayram F, Kurtoglu S. Cross-sectional reference values for mid-upper arm circumference, triceps skinfold thickness and arm fat area of Turkish children and adolescents. International journal of food sciences and nutrition. 2009;60(4):267-281. 161. Ayatollahi S, Shayan Z. New reference values for mid upper arm circumference of Shiraz (Iran) primary school children. e-SPEN, the European e-Journal of Clinical Nutrition and Metabolism. 2008;3(2):e72-e77. 162. Wells J, Fewtrell M. Measuring body composition. Archives of Disease in Childhood. 2006;91(7):612-617. 163. Wells JC, Sharp G, Steer PJ, Leon DA. Paternal and maternal influences on differences in birth weight between Europeans and Indians born in the UK. 2013. 164. Bank TW. Scaling up School Feeding: keeping children in school while improving their learning and health. 2012. 165. Galloway R, Kristjansson E, Gelli A, Meir U, Espejo F, Bundy D. School feeding: Outcomes and costs. Food & Nutrition Bulletin. 2009;30(2):171-182. 166. Grantham-McGregor SM, Chang S, Walker SP. Evaluation of school feeding programs: some Jamaican examples. The American journal of clinical nutrition. 1998;67(4):785S- 789S. 167. Kazianga H, De Walque D, Alderman H. Educational and health impacts of two school feeding schemes: Evidence from a randomized trial in rural Burkina Faso. World Bank Policy Research Working Paper. 2009(4976). 168. Programme WF. Home-grown School Feeding: A Framework to Link School Feeding with Local Agricultural Production. World Food Programme; 2009. 169. Bishai D, Nalubola R. The History of Food Fortification in the United States: Its Relevance for Current Fortification Efforts in Developing Countries*. Economic Development and Cultural Change. 2002;51(1):37-53. 170. Park YK, Sempos CT, Barton CN, Vanderveen JE, Yetley EA. Effectiveness of food fortification in the United States: the case of pellagra. American journal of public health. 2000;90(5):727. 171. Allen LH, De Benoist B, Dary O, Hurrell R. Guidelines on food fortification with micronutrients. 2006. 172. Self JL, Serdula M, Dowswell T, De-Regil LM. Fortification of condiments and seasonings with iron for preventing anaemia and improving health. The Cochrane Library. 2012.

181

173. Piccoli NB, Grede N, de Pee S, et al. Rice fortification: its potential for improving micronutrient intake and steps required for implementation at scale. Food & Nutrition Bulletin. 2012;33(Supplement 3):360S-372S. 174. Roks E. Review of the cost components of introducing industrially fortified rice. Annals of the New York Academy of Sciences. 2014;1324(1):82-91. 175. Radhika MS, Nair KM, Kumar RH, et al. Micronized ferric pyrophosphate supplied through extruded rice kernels improves body iron stores in children: a double-blind, randomized, placebo-controlled midday meal feeding trial in Indian schoolchildren. Am J Clin Nutr. 2011;94(5):1202-1210. 176. Moretti D, Zimmermann MB, Muthayya S, et al. Extruded rice fortified with micronized ground ferric pyrophosphate reduces iron deficiency in Indian schoolchildren: a double- blind randomized controlled trial. Am J Clin Nutr. 2006;84(4):822-829. 177. Pinkaew S, Winichagoon P, Hurrell RF, Wegmuller R. Extruded rice grains fortified with zinc, iron, and vitamin A increase zinc status of Thai school children when incorporated into a school lunch program. The Journal of nutrition. 2013;143(3):362-368. 178. Alavi S, Bugusu B, Cramer G, et al. Rice fortification in developing countries: a critical review of the technical and economic feasibility. Institute of Food Technologists, Washington DC. 2008. 179. Khanh Van T, Burja K, Thuy Nga T, et al. Organoleptic qualities and acceptability of fortified rice in two Southeast Asian countries. Annals of the New York Academy of Sciences. 2014;1324(1):48-54. 180. Pee S. Proposing nutrients and nutrient levels for rice fortification. Annals of the New York Academy of Sciences. 2014;1324(1):55-66. 181. Meltzer HM, Aro A, Andersen NL, Koch B, Alexander J. Risk analysis applied to food fortification. Public health nutrition. 2003;6(03):281-290. 182. Kraemer K, Waelti M, De Pee S, et al. Are low tolerable upper intake levels for vitamin A undermining effective food fortification efforts? Nutrition reviews. 2008;66(9):517-525. 183. FAO. The State of Food Insecurity in the World. Adressing food insecurity in protracted crises. : FAO;2010. 184. UNDP. Human Development Report 2011. New York: UNDP;2011. 185. FAO. Profil nutritionnel du Sénégal. 2010. 186. WFP/UNICEF/FAO/ANSD. Impact de la hausse des prix sur la sécurité alimentaire des populations en milieu urbain. Sénégal : Pikine-Kaolack-Zinguichor. WFP; 2008. 187. Becquey E, Martin-Prevel Y. Micronutrient adequacy of women's diet in urban Burkina Faso is low. J Nutr. Vol 140. United States2010:2079S-2085S. 188. Gordon RC, Rose MC, Skeaff SA, Gray AR, Morgan KM, Ruffman T. Iodine supplementation improves cognition in mildly iodine-deficient children. Am J Clin Nutr. 2009;90(5):1264-1271. 189. Black RE, Sazawal S. Zinc and childhood infectious disease morbidity and mortality. Br J Nutr. Vol 85 Suppl 2. England2001:S125-129. 190. Pandey A, Chakraborty AK. Undernutrition, vitamin A deficiency and ARI morbidity in underfives. Indian J Public Health. 1996;40(1):13-16. 191. Gueye. Anémie et état nutritionnel d'enfants d'âge scolaire du CI au CE2 de l'élémentaire public du département de Dakar. Mémoire de DEA : UCAD; 1994. 192. ANSD. Situation économique et sociale de la région de Dakar de l'année 2008. 2009. 193. Kaiser R, Woodruff BA, Bilukha O, Spiegel PB, Salama P. Using design effects from previous cluster surveys to guide sample size. Disasters. 2006;30(2):199-211. 194. Cole TJ, Flegal KM, Nicholls D, Jackson AA. Body mass index cut offs to define thinness in children and adolescents. Bmj. 2007;335(7612):194. 195. Erhardt JG, Estes JE, Pfeiffer CM, Biesalski HK, Craft NE. Combined measurement of ferritin, soluble transferrin receptor, retinol binding protein, and C-reactive protein by an inexpensive, sensitive, and simple sandwich enzyme-linked immunosorbent assay technique. J Nutr. 2004;134(11):3127-3132.

182

196. Pino S, Fang SL, Braverman LE. Ammonium persulfate: a safe alternative oxidizing reagent for measuring urinary iodine. Clin Chem. 1996;42(2):239-243. 197. Thurnham DI, McCabe LD, Haldar S, Wieringa FT, Northrop-Clewes CA, McCabe GP. Adjusting plasma ferritin concentrations to remove the effects of subclinical. Am J Clin Nutr. 2010;92(3):546-555. 198. Organization WH. Assessing the iron status of populations: including literature reviews: report of a Joint World Health Organization/Centers for Disease Control and Prevention Technical Consultation on the Assessment of Iron Status at the Population Level, Geneva, Switzerland, 6 –8 April 2004. Geneva: by WHO/CDC. 2007. 199. Cook JD, Flowers CH, Skikne BS. The quantitative assessment of body iron. Blood. 2003;101(9):3359-3363. 200. de Pee S, Dary O. Biochemical indicators of vitamin A deficiency: Serum retinol and serum retinol binding protein. Journal of Nutrition. 2002;132(9):2895S-2901S. 201. Thurnham DI, McCabe GP, Northrop-Clewes CA, Nestel P. Effects of subclinical infection on plasma retinol concentrations and assessment of prevalence of vitamin A deficiency: meta-analysis. Lancet. 2003;362(9401):2052-2058. 202. Gibson RS. Principles of nutritional assessment. New York: Oxford press; 1990:378-388. 203. Brown KH, Rivera JA, Bhutta Z, et al. International Zinc Nutrition Consultative Group (IZiNCG) technical document #1. Food Nutr Bull. 2004;25(1 Suppl 2):S99-203. 204. Bovet P, Kizirian N, Madeleine G, Blossner M, Chiolero A. Prevalence of thinness in children and adolescents in the Seychelles: comparison. Nutr J. 2011;10:65. 205. UNICEF. Senegal statistics. HIV/AIDS 2011; http://www.unicef.org/infobycountry/senegal_statistics.html . Available at, 2013. 206. WHO. Global Tuberculosis Report. Geneva: World Health Organization;2012. 207. Martorell R. The nature of child malnutrition and its long-term implications. Food and Nutrition Bulletin. 1999;20(3). 208. de Benoist B, McLean E, Egli I, Cogswell M. Worldwide prevalence of anaemia 1993 - 2005 : WHO global database on anaemia. WHO;2008. 209. Idohou-Dossou N, Wade S, Guiro AT, et al. Nutritional status of preschool Senegalese children: long-term effects of early. Br J Nutr. 2003;90(6):1123-1132. 210. Zimmermann MB, Molinari L, Staubli-Asobayire F, et al. Serum transferrin receptor and zinc protoporphyrin as indicators of iron status. Am J Clin Nutr. 2005;81(3):615-623. 211. El Hioui M, Ahami AO, Aboussaleh Y, et al. Risk factors of anaemia among rural school children in Kenitra, Morocco. East Afr J Public Health. 2008;5(2):62-66. 212. PNLP. Focus sur le Sénégal. Collection progrès et impact. Vol 4: Roll back malaria; 2010. 213. Diagne I, Diagne-Gueye NDR, Signate-Sy H, et al. Prise en charge de la drépanocytose chez l'enfant en Afrique : experience de la cohorte de l'hopital d'enfants Albert Royer de Dakar. 2003. 214. Dreyfuss ML, Stoltzfus RJ, Shrestha JB, et al. Hookworms, malaria and vitamin A deficiency contribute to anemia and iron deficiency among pregnant women in the plains of Nepal. J Nutr. 2000;130(10):2527-2536. 215. Dabone C, Delisle HF, Receveur O. Poor nutritional status of schoolchildren in urban and peri-urban areas of. Nutr J. 2011;10:34. 216. Kassaye T, Receveur O, Johns T, Becklake MR. Prevalence of vitamin A deficiency in children aged 6-9 years in Wukro, northern. Bull World Health Organ. 2001;79(5):415- 422. 217. Baeten JM, Richardson BA, Bankson DD, et al. Use of serum retinol-binding protein for prediction of vitamin A deficiency. Am J Clin Nutr. 2004;79(2):218-225. 218. Wahed MA, Alvarez JO, Khaled MA, Mahalanabis D, Rahman MM, Habte D. Comparison of the modified relative dose response (MRDR) and the relative dose. Am J Clin Nutr. 1995;61(6):1253-1256. 219. Wieringa FT, Berger J, Dijkhuizen MA, et al. Sex differences in prevalence of anaemia and iron deficiency in infancy in a large multi-country trial in South-East Asia. Br J Nutr. 2007;98(5):1070-1076.

183

220. Cavan KR, Gibson RS, Grazioso CF, Isalgue AM, Ruz M, Solomons NW. Growth and body composition of periurban Guatemalan children in relation to zinc. Am J Clin Nutr. 1993;57(3):334-343. 221. Qin Y, Melse-Boonstra A, Zhao J, Wu M, Hu X, Kok FJ. Stunting and zinc deficiency among primary school children in rural areas with. Asia Pac J Clin Nutr. 2009;18(1):15-21. 222. Gibson RS, Manger MS, Krittaphol W, et al. Does zinc deficiency play a role in stunting among primary school children in NE. Br J Nutr. 2007;97(1):167-175. 223. Thurlow RA, Winichagoon P, Pongcharoen T, et al. Risk of zinc, iodine and other micronutrient deficiencies among school children. Eur J Clin Nutr. 2006;60(5):623-632. 224. Dapi N, Janlert U, Nouedoui C, Håglin L. Adolescents´ food habits and nutritional status in urban and rural areas in Cameroon, Africa. Scandinavian Journal of Nutrition/Næringsforskning. 2005;49(4):151-158. 225. Fiorentino M, Bastard G, Sembène M, et al. Anthropometric and micronutrient status of school-children in an urban West Africa setting: a cross-sectional study in dakar (senegal). PloS one. 2013;8(12):e84328. 226. Preziosi P, Deheeger M. Portions alimentaires: manuel-photos pour l'estimation des quantités. Polytechnica; 2002. 227. Stadlmayr B, Charrondiere UR, Addy P, et al. Composition of selected foods from West Africa. Food and Agriculture Organization, Rome. 2010:13-14. 228. Travail AndssAE. French food composition table CIQUAL 2008. 2008; http://www.afssa.fr/TableCIQUAL , 2011. 229. Bunch SM, Suzanne. User's guide to the operation of the World Food Dietary Assessment Program. Office of Technology Licensing, University of California, Berkeley, USA1997. 230. Gebhardt S, Lemar L, Haytowitz D, et al. USDA national nutrient database for standard reference, release 21. 2008. 231. Guenther PM, Kott PS, Carriquiry AL. Development of an approach for estimating usual nutrient intake distributions at the population level. The Journal of nutrition. 1997;127(6):1106-1112. 232. Micronutrients IoMPo. Dietary Reference Intakes for Energy, Carbohydrate, Fiber, Fat, Fatty Acids, Cholesterol, Protein and Amino Acids. Panel on Macronutrients Panel on the Definition of Dietary Fiber, Subcommittee on Upper Reference Levels of Nutrients, Subcommittee on Interpretation and Uses of Dietary Reference Intakes, and the Standing Committee on the Scientific Evaluation of Dietary Reference Intakes, Food and Nutrition Board. National Academies Press; 2005. 233. Board IoMFaN. DRI, Dietary Reference Intakes for Vitamin A, Vitamin K, Arsenic, Boron, Chromium, Copper, Iodine, Iron, Manganese, Molybdenum, Nickel, Silicon, Vanadium, and Zinc: A Report of the Panel on Micronutrients and of Interpretation and Uses of Dietary Reference Intakes, and the Standing Committee on the Scientific Evaluation of Dietary Reference Intakes. National Academy Press; 2001. 234. Uauy R, Dangour AD. Fat and fatty acid requirements and recommendations for infants of 0-2 years and children of 2-18 years. Ann Nutr Metab. 2009;55(1-3):76-96. 235. Carriquiry AL. Assessing the prevalence of nutrient inadequacy. Public health nutrition. 1999;2(01):23-34. 236. Dapi LN, Hörnell A, Janlert U, Stenlund H, Larsson C. Energy and nutrient intakes in relation to sex and socio-economic status among school adolescents in urban Cameroon, Africa. Public health nutrition. 2011;14(05):904-913. 237. Abizari AR, Buxton C, Kwara L, Mensah-Homiah J, Armar-Klemesu M, Brouwer ID. School feeding contributes to micronutrient adequacy of Ghanaian schoolchildren. Br J Nutr. 2014;112(6):1019-1033. 238. Hellwig JP, Otten JJ, Meyers LD. Dietary Reference Intakes:: The Essential Guide to Nutrient Requirements. National Academies Press; 2006. 239. Bingham SA, Day NE, Luben R, et al. Dietary fibre in food and protection against colorectal cancer in the European Prospective Investigation into Cancer and Nutrition (EPIC): an observational study. The lancet. 2003;361(9368):1496-1501.

184

240. Rose DJ, DeMeo MT, Keshavarzian A, Hamaker BR. Influence of dietary fiber on inflammatory bowel disease and colon cancer: importance of fermentation pattern. Nutrition reviews. 2007;65(2):51-62. 241. Järvinen R, Knekt P, Hakulinen T, Rissanen H, Heliövaara M. Dietary fat, cholesterol and colorectal cancer in a prospective study. British journal of cancer. 2001;85(3):357. 242. Pollitt E. Functional significance of the covariance between protein energy malnutrition and iron deficiency anemia. The Journal of nutrition. 1995;125(8):2272S. 243. Eder K, Kirchgessner M. Dietary fat influences the effect of zinc deficiency on liver lipids and fatty acids in rats force-fed equal quantities of diet. J Nutr. 1994;124(10):1917- 1926. 244. Smuts C, Tichelaar H, Van Jaarsveld P, et al. The effect of iron fortification on the fatty acid composition of plasma and erythrocyte membranes in primary school children with and without iron deficiency. Prostaglandins, leukotrienes and essential fatty acids. 1995;52(1):59-67. 245. Gewa CA, Murphy SP, Weiss RE, Neumann CG. Determining minimum food intake amounts for diet diversity scores to maximize associations with nutrient adequacy: an analysis of schoolchildren's diets in rural Kenya. Public Health Nutr. 2014;17(12):2667- 2673. 246. Zimmermann MB, Chaouki N, Hurrell RF. Iron deficiency due to consumption of a habitual diet low in bioavailable iron: a longitudinal cohort study in Moroccan children. Am J Clin Nutr. 2005;81(1):115-121. 247. Samuelson G, Bratteby LE, Berggren K, Elverby JE, Kempe B. Dietary iron intake and iron status in adolescents. Acta paediatrica. 1996;85(9):1033-1038. 248. Jang J-T, Green JB, Beard JL, Green MH. Kinetic analysis shows that iron deficiency decreases liver vitamin A mobilization in rats. The Journal of nutrition. 2000;130(5):1291-1296. 249. Opoku Danquah ANA, Adwoa. Opare-Obisaw, Clara. Nutritional status of upper primary school pupils in a rural setting in Ghana2013. 250. Jacques PF, Sulsky SI, Sadowski JA, Phillips JC, Rush D, Willett WC. Comparison of micronutrient intake measured by a dietary questionnaire and biochemical indicators of micronutrient status. Am J Clin Nutr. 1993;57(2):182-189. 251. l'alimentation Cndcdéerslne, scientifique Cndlr, aliments Afdssd. Apports nutritionnels conseillés pour la population française. Tec & doc; 2000. 252. Group ICoMRE. Nutrient Requirements and Recommended Dietary Allowances for Indians: A Report of the Expert Group of the Indian Council of Medical Research. Indian Council of Medical Research; 1990. 253. Anzid K, Baali A, Vimard P, Levy-Desroches S, Cherkaoui M, López PM. Inadequacy of vitamins and minerals among high-school pupils in Ouarzazate, Morocco. Public health nutrition. 2014;17(08):1786-1795. 254. Eyberg CJ, Pettifor J, Moodley G. Dietary calcium intake in rural black South African children. The relationship between calcium intake and calcium nutritional status. Human nutrition. Clinical nutrition. 1986;40(1):69-74. 255. Uush T. Calcium intake and serum calcium status in Mongolian children. The Journal of steroid biochemistry and molecular biology. 2014;144:167-171. 256. Pettifor JM. Nutritional rickets: deficiency of vitamin D, calcium, or both? The American journal of clinical nutrition. 2004;80(6):1725S-1729S. 257. Balasubramanian K, Rajeswari J, Govil Y, Agarwal A, Kumar A, Bhatia V. Varying role of vitamin D deficiency in the etiology of rickets in young children vs. adolescents in northern India. Journal of tropical pediatrics. 2003;49(4):201-206. 258. Al-Jurayyan NA, El-Desouki ME, Al-Herbish AS, Al-Mazyad AS, Al-Qhtani MM. Nutritional rickets and osteomalacia in school children and adolescents. Saudi medical journal. 2002;23(2):182-185. 259. Pettifor JM. Vitamin D &/or calcium deficiency rickets in infants & children: a global perspective. Indian Journal of Medical Research. 2008;127(3):245.

185

260. Broutin C, Duteurtre V, Tandia A, Touré B, François M. Accroissement et diversification de l'offre de produits laitiers au Sénégal: la bataille industrielle du lait en poudre à Dakar et des minilaiteries à la conquête des marchés des villes secondaires. Revue d'élevage et de médecine vétérinaire des pays tropicaux. 2007;60(1-4). 261. Ene-Obong HN, Odoh IF, Ikwuagwu OE. Plasma vitamin A and C status of in-school adolescents and associated factors in Enugu State, Nigeria. J Health Popul Nutr. 2003;21(1):18-25. 262. Byers T, Treiber F, Gunter E, et al. The accuracy of parental reports of their children's intake of fruits and vegetables: validation of a food frequency questionnaire with serum levels of carotenoids and vitamins C, A, and E. Epidemiology. 1993;4(4):350-355. 263. De Pee S, West C, Permaesih D, Martuti S, Hautvast J. Orange fruit is more effective than are dark-green, leafy vegetables in increasing serum concentrations of retinol and beta- carotene in schoolchildren in Indonesia. The American journal of clinical nutrition. 1998;68(5):1058-1067. 264. Naidu KA. Vitamin C in human health and disease is still a mystery? An overview. Nutrition Journal. 2003;2(1):7. 265. Padayatty SJ, Katz A, Wang Y, et al. Vitamin C as an antioxidant: evaluation of its role in disease prevention. Journal of the American College of Nutrition. 2003;22(1):18-35. 266. Holt EM, Steffen LM, Moran A, et al. Fruit and vegetable consumption and its relation to markers of inflammation and oxidative stress in adolescents. Journal of the American Dietetic Association. 2009;109(3):414-421. 267. Wintergerst ES, Maggini S, Hornig DH. Immune-enhancing role of vitamin C and zinc and effect on clinical conditions. Annals of Nutrition and Metabolism. 2006;50(2):85-94. 268. Cook JD, Reddy MB. Effect of ascorbic acid intake on nonheme-iron absorption from a complete diet. The American journal of clinical nutrition. 2001;73(1):93-98. 269. Lane DJ, Richardson DR. The active role of vitamin C in mammalian iron metabolism: much more than just enhanced iron absorption! Free Radic Biol Med. 2014;75:69-83. 270. Lynch SR, Cook JD. Interaction of vitamin C and iron. Ann NY Acad Sci. 1980;355(3). 271. Wang X, Seib P, Ra K. L-Ascorbic Acid and Its 2-Phosphorylated Derivatives in Selected Foods: Vitamin C Fortification and Antioxidant Properties. Journal of food science. 1995;60(6):1295-1300. 272. Payette H, Gray-Donald K. Dietary intake and biochemical indices of nutritional status in an elderly population, with estimates of the precision of the 7-d food record. The American journal of clinical nutrition. 1991;54(3):478-488. 273. Bates CJ, Prentice AM, Paul A. Seasonal variations in vitamins A, C, riboflavin and folate intakes and status of pregnant and lactating women in a rural Gambian community: some possible implications. European journal of clinical nutrition. 1994;48(9):660-668. 274. Gilbert CE, Waddel K, Wood M, Foster A. Causes of childhood blindness in east Africa: results in 491 pupils attending 17 schools for the blind in Malawi, Kenya and Uganda. Ophthalmic epidemiology. 1995;2(2):77-84. 275. Webb P. Intrahousehold dimensions of micronutrient deficiencies: a review of the evidence. Friedman School of Nutrition Science and Policy;2002. 276. UNICEF. UNICEF Annual Report 2013 – Senegal. 2013. 277. Matheson DM, Varady J, Varady A, Killen JD. Household food security and nutritional status of Hispanic children in the fifth grade. The American journal of clinical nutrition. 2002;76(1):210-217. 278. Allen LH. Causes of vitamin B12 and folate deficiency. Food & Nutrition Bulletin. 2008;29(Supplement 1):20-34. 279. Meme M, Kogi-Makau W, Muroki N, Mwadime R. Energy and protein intake and nutritional status of primary schoolchildren 5 to 10 years of age in schools with and without feeding programmes in Nyambene District, Kenya. Food & Nutrition Bulletin. 1998;19(4):334-342.

186

280. Perignon M, Fiorentino M, Kuong K, et al. Multi-micronutrient fortified rice improved vitamin A status of Cambodian school children. Micronutrient Forum Global Conference; 2014; Addis Ababa, Ethiopia. 281. Dibba B, Prentice A, Ceesay M, Stirling DM, Cole TJ, Poskitt EM. Effect of calcium supplementation on bone mineral accretion in Gambian children accustomed to a low- calcium diet. The American journal of clinical nutrition. 2000;71(2):544-549. 282. Lee W, Leung S, Wang S-H, et al. Double-blind, controlled calcium supplementation and bone mineral accretion in children accustomed to a low-calcium diet. The American journal of clinical nutrition. 1994;60(5):744-750. 283. Bonjour J-P, Chevalley T, Ammann P, Slosman D, Rizzoli R. Gain in bone mineral mass in prepubertal girls 3 –5 years after discontinuation of calcium supplementation: a follow- up study. The Lancet. 2001;358(9289):1208-1212. 284. Bingham SA, Gill C, Welch A, et al. Validation of dietary assessment methods in the UK arm of EPIC using weighed records, and 24-hour urinary nitrogen and potassium and serum vitamin C and carotenoids as biomarkers. International journal of epidemiology. 1997;26(suppl 1):S137. 285. McPherson RS, Hoelscher DM, Alexander M, Scanlon KS, Serdula MK. Dietary assessment methods among school-aged children: validity and reliability. Preventive Medicine. 2000;31(2):S11-S33. 286. Brussaard J, Löwik M, Steingrimsdottir L, et al. A European food consumption survey method--conclusions and recommendations. European Journal of Clinical Nutrition. 2002;56:S89-94. 287. Willett W. Nutritional epidemiology. Oxford University Press; 2012. 288. Livingstone M, Robson P, Wallace J. Issues in dietary intake assessment of children and adolescents. British Journal of Nutrition. 2004;92(S2):S213-S222. 289. Gewa CA, Murphy SP, Neumann CG. Out-of-home food intake is often omitted from mothers' recalls of school children's intake in rural Kenya. The Journal of nutrition. 2007;137(9):2154-2159. 290. Dop M-C, Milan C, Milan C, N'Diaye A. The 24-hour recall for Senegalese weanlings: a validation exercise. European journal of clinical nutrition. 1994;48(9):643-653. 291. Ribas-Barba L, Serra-Majem L, Román-Vinas B, Ngo J, García-Álvarez A. Effects of dietary assessment methods on assessing risk of nutrient intake adequacy at the population level: from theory to practice. British Journal of Nutrition. 2009;101(S2):S64-S72. 292. De Lauzon B, Volatier J, Martin A. A Monte Carlo simulation to validate the EAR cut-point method for assessing the prevalence of nutrient inadequacy at the population level. Public health nutrition. 2004;7(07):893-900. 293. Murphy SP, Poos MI. Dietary reference intakes: summary of applications in dietary assessment. Public health nutrition. 2002;5(6a):843-849. 294. Román-Viñas B, Serra-Majem L, Ribas-Barba L, et al. Overview of methods used to evaluate the adequacy of nutrient intakes for individuals and populations. British Journal of Nutrition. 2009;101(S2):S6-S11. 295. National Institute of Statistics DGfHaIM. Cambodia Demographic and Health Survey 2010. Phnom Penh, Cambodia and Calverton, Maryland, USA2011. 296. UNICEF MI. Vitamin and Mineral Deficiency - A Global Progress Report. 2004. 297. Falkingham M, Abdelhamid A, Curtis P, Fairweather-Tait S, Dye L, Hooper L. The effects of oral iron supplementation on cognition in older children and adults: a systematic review and meta-analysis. Nutr J. 2010;9:4. 298. Khor GL, Misra S. Micronutrient interventions on cognitive performance of children aged 5-15 years in developing countries. Asia Pac J Clin Nutr. 2012;21(4):476-486. 299. Walker SP, Wachs TD, Gardner JM, et al. Child development: risk factors for adverse outcomes in developing countries. Lancet. 2007;369(9556):145-157. 300. Black RE, Allen LH, Bhutta ZA, et al. Maternal and child undernutrition 1 - Maternal and child undernutrition: global and regional exposures and health consequences. Lancet. 2008;371(9608):243-260.

187

301. Crookston BT, Dearden KA, Alder SC, et al. Impact of early and concurrent stunting on cognition. Matern Child Nutr. 2011;7(4):397-409. 302. Grantham-McGregor S. Linear growth retardation and cognition. The Lancet. 2002;359(9306):542. 303. Sokolovic N, Selvam S, Srinivasan K, Thankachan P, Kurpad AV, Thomas T. Catch-up growth does not associate with cognitive development in Indian school-age children. European Journal of Clinical Nutrition. 2014;68(1):14-18. 304. Sandjaja, Poh BK, Rojroonwasinkul N, et al. Relationship between anthropometric indicators and cognitive performance in Southeast Asian school-aged children. British Journal of Nutrition. 2013;110:S57-S64. 305. Sowell ER, Thompson PM, Toga AW. Mapping changes in the human cortex throughout the span of life. Neuroscientist. 2004;10(4):372-392. 306. Kaplan RM, Saccuzzo DP. Psychological Testing: Principles, Applications, and Issues: Principles, Applications, and Issues. Wadsworth Cengage Learning; 2009. 307. Hughes D, Bryan J. The assessment of cognitive performance in children: considerations for detecting nutritional influences. Nutr Rev. 2003;61(12):413-422. 308. WHO. Iron Deficiency Anaemia Assessment, Prevention and Control. A guide for programme managers. Geneva: World Health Organization;2001. 309. Skikne BS, Flowers CH, Cook JD. Serum transferrin receptor: a quantitative measure of tissue iron deficiency. Blood. 1990;75(9):1870-1876. 310. Gibson RS. Principles of Nutritional Assessment. Oxford University Press; 2005. 311. Gibson RS, Hess SY, Hotz C, Brown KH. Indicators of zinc status at the population level: a review of the evidence. British Journal of Nutrition. 2008;99:S14-S23. 312. Dunn JT, Crutchfield HE, Gutekunst R, Dunn AD. Methods for measuring iodine in urine. ICCIDD, UNICEF, WHO; 1993. 313. WHO. Assessment of iodine deficiency disorders and monitoring their elimination. A guide for programme managers. 3rd ed. World Health Organization2007:32-33. 314. Ash LR, Savioli L, Orihel TC, World Health O. Bench aids for the diagnosis of intestinal parasites. Geneva: World Health Organization; 1994. 315. Zimmermann MB. Methods to assess iron and iodine status. British Journal of Nutrition. 2008;99:S2-S9. 316. Knowles J, Thurnham DI, Phengdy B, et al. Impact of inflammation on the biomarkers of iron status in a cross-sectional survey of Lao women and children. Br J Nutr. 2013;110(12):2285-2297. 317. Victora CG, de Onis M, Hallal PC, Blossner M, Shrimpton R. Worldwide timing of growth faltering: revisiting implications for interventions. Pediatrics. 2010;125(3):e473-480. 318. National Institute of Public Health, National Institute of Statistics, Macro O. Cambodia Demographic and Health Survey 2005. 2006. 319. Hall A. Micronutrient supplements for children after deworming. The Lancet Infectious Diseases. 2007;7(4):297-302. 320. Hall A, Hewitt G, Tuffrey V, de Silva N. A review and meta-analysis of the impact of intestinal worms on child growth and nutrition. Matern Child Nutr. 2008;4 Suppl 1:118- 236. 321. Use and Interpretation of Anthropometric Indicators of Nutritional-Status. Bulletin of the World Health Organization. 1986;64(6):929-941. 322. Chandra RK. Immune responses in parasitic diseases. Part B: mechanisms. Rev Infect Dis. 1982;4(4):756-762. 323. Bundy DA, Golden MH. The impact of host nutrition on gastrointestinal helminth populations. Parasitology. 1987;95 ( Pt 3):623-635. 324. World Health Organization., Food and Agriculture Organization of the United Nations., International Atomic Energy Agency. Trace elements in human nutrition and health. Geneva: World Health Organization; 1996. 325. Fischer Walker CL, Black RE. Functional indicators for assessing zinc deficiency. Food Nutr Bull. 2007;28(3 Suppl):S454-479.

188

326. Gibson RS. A historical review of progress in the assessment of dietary zinc intake as an indicator of population zinc status. Adv Nutr. 2012;3(6):772-782. 327. Hotez PJ, Kamath A. Neglected tropical diseases in sub-saharan Africa: review of their prevalence, distribution, and disease burden. PLoS Negl Trop Dis. 2009;3(8):e412. 328. Ahmed A, Al-Mekhlafi HM, Al-Adhroey AH, Ithoi I, Abdulsalam AM, Surin J. The nutritional impacts of soil-transmitted helminths infections among Orang Asli schoolchildren in rural Malaysia. Parasit Vectors. 2012;5:119. 329. Bethony J, Brooker S, Albonico M, et al. Soil-transmitted helminth infections: ascariasis, trichuriasis, and hookworm. Lancet. 2006;367(9521):1521-1532. 330. Rajagopal S, Hotez PJ, Bundy DA. Micronutrient supplementation and deworming in children with geohelminth infections. PLoS Negl Trop Dis. 2014;8(8):e2920. 331. Kongsbak K, Wahed MA, Friis H, Thilsted SH. Acute phase protein levels, T. trichiura, and maternal education are predictors of serum zinc in a cross-sectional study in Bangladeshi children. J Nutr. 2006;136(8):2262-2268. 332. Koski KG, Scott ME. Gastrointestinal nematodes, nutrition and immunity: breaking the negative spiral. Annu Rev Nutr. 2001;21:297-321. 333. Sinuon M, Anantaphruti MT, Socheat D. Intestinal helminthic infections in schoolchildren in Cambodia. Southeast Asian J Trop Med Public Health. 2003;34(2):254- 258. 334. Franco M, Bilal U, Ordunez P, et al. Population-wide weight loss and regain in relation to diabetes burden and cardiovascular mortality in Cuba 1980-2010: repeated cross sectional surveys and ecological comparison of secular trends. BMJ. 2013;346:f1515. 335. Wordemann M, Polman K, Menocal Heredia LT, et al. Prevalence and risk factors of intestinal parasites in Cuban children. Trop Med Int Health. 2006;11(12):1813-1820. 336. Katz N, Chaves A, Pellegrino J. A simple device for quantitative stool thick-smear technique in Schistosomiasis mansoni. Rev Inst Med Trop Sao Paulo. 1972;14(6):397- 400. 337. Nunez Fernández FA, Sanjurjo González E, Finlay CM, Gálvez Oviedo D. Estudio de dosis única de Mebendazol, para tratamiento de Trichuris trichiura y Necator americanus en las comunidades. Rev. Cuba. de Med. Trop. 1989;41:371 –378. 338. Vercruysse J, Behnke JM, Albonico M, et al. Assessment of the anthelmintic efficacy of albendazole in school children in seven countries where soil-transmitted helminths are endemic. PLoS Negl Trop Dis. 2011;5(3):e948. 339. Wieringa FT, Dijkhuizen MA, West CE, Northrop-Clewes CA. Estimation of the effect of the acute phase response on indicators of micronutrient status in Indonesian infants. J Nutr. 2002;132(10):3061-3066. 340. D'Haese PC, Lamberts LV, Vanheule AO, De Broe ME. Direct determination of zinc in serum by Zeeman atomic absorption spectrometry with a graphite furnace. Clin Chem. 1992;38(12):2439-2443. 341. Hambidge KM, Hambidge C, Jacobs M, Baum JD. Low levels of zinc in hair, anorexia, poor growth, and hypogeusia in children. Pediatr Res. 1972;6(12):868-874. 342. Rosado JL, Caamano MC, Montoya YA, de Lourdes Solano M, Santos JI, Long KZ. Interaction of zinc or vitamin A supplementation and specific parasite infections on Mexican infants' growth: a randomized clinical trial. Eur J Clin Nutr. 2009;63(10):1176- 1184. 343. Osei A, Houser R, Bulusu S, Joshi T, Hamer D. Nutritional status of primary schoolchildren in Garhwali Himalayan villages of India. Food Nutr Bull. 2010;31(2):221- 233. 344. de Gier B, Campos Ponce M, van de Bor M, Doak CM, Polman K. Helminth infections and micronutrients in school-age children: a systematic review and meta-analysis. Am J Clin Nutr. 2014;99(6):1499-1509. 345. Yap P, Utzinger J, Hattendorf J, Steinmann P. Influence of nutrition on infection and re- infection with soil-transmitted helminths: a systematic review. Parasit Vectors. 2014;7:229.

189

346. Papier K, Williams GM, Luceres-Catubig R, et al. Childhood malnutrition and parasitic helminth interactions. Clin Infect Dis. 2014;59(2):234-243. 347. Grantham-McGregor S, Cheung YB, Cueto S, et al. Developmental potential in the first 5 years for children in developing countries. Lancet. 2007;369(9555):60-70. 348. Korpe PS, Petri WA, Jr. Environmental enteropathy: critical implications of a poorly understood condition. Trends Mol Med. 2012;18(6):328-336. 349. Moshage H. Cytokines and the hepatic acute phase response. J Pathol. 1997;181(3):257- 266. 350. Mayo-Wilson E, Imdad A, Junior J, Dean S, Bhutta ZA. Preventive zinc supplementation for children, and the effect of additional iron: a systematic review and meta-analysis. BMJ Open. 2014;4(6):e004647. 351. Lowe NM, Fekete K, Decsi T. Methods of assessment of zinc status in humans: a systematic review. Am J Clin Nutr. 2009;89(6):2040S-2051S. 352. Golden MH. Proposed recommended nutrient densities for moderately malnourished children. Food Nutr Bull. 2009;30(3 Suppl):S267-342. 353. Thurnham DI. Interactions between nutrition and immune function: using inflammation biomarkers to interpret micronutrient status. Proc Nutr Soc. 2014;73(1):1-8. 354. Organization WH. The global prevalence of anaemia in 2011. Geneva Switzerland: WHO 2015; 2015. 355. Marengo-Rowe AJ. The thalassemias and related disorders. Proceedings (Baylor University. Medical Center). 2007;20(1):27. 356. Fucharoen S, Winichagoon P. Haemoglobinopathies in southeast Asia. The Indian journal of medical research. 2011;134(4):498. 357. Best C, Neufingerl N, Del Rosso JM, Transler C, van den Briel T, Osendarp S. Can multi- micronutrient food fortification improve the micronutrient status, growth, health, and cognition of schoolchildren? A systematic review (PAYANT). Nutr Rev. 2011;69(4):186- 204. 358. Consensus C. http://www.copenhagenconsensus.com/copenhagen-consensus- ii/outcomes , 2015. 359. Beinner MA, Velasquez-Melendez G, Pessoa MC, Greiner T. Iron-fortified rice is as efficacious as supplemental iron drops in infants and young children. J Nutr. 2010;140(1):49-53. 360. Hotz C, Porcayo M, Onofre G, et al. Efficacy of iron-fortified Ultra Rice in improving the iron status of women in Mexico. Food Nutr Bull. 2008;29(2):140-149. 361. Nga TT, Winichagoon P, Dijkhuizen MA, et al. Multi-Micronutrient –Fortified Biscuits Decreased Prevalence of Anemia and Improved Micronutrient Status and Effectiveness of Deworming in Rural Vietnamese School Children. The Journal of nutrition. 2009;139(5):1013-1021. 362. Skikne BS. Serum transferrin receptor. American journal of hematology. 2008;83(11):872-875. 363. Harlow LL. The Essence of Multivariate Thinking: Basic Themes and Methods: Basic Themes and Methods. Routledge; 2014. 364. Hix J, Rasca P, Morgan J, et al. Validation of a rapid enzyme immunoassay for the quantitation of retinol-binding protein to assess vitamin A status within populations. European journal of clinical nutrition. 2006;60(11):1299-1303. 365. Zimmermann MB, Fucharoen S, Winichagoon P, et al. Iron metabolism in heterozygotes for hemoglobin E (HbE), α -thalassemia 1, or β -thalassemia and in compound heterozygotes for HbE/β -thalassemia. The American journal of clinical nutrition. 2008;88(4):1026-1031. 366. Andrews NC. Anemia of inflammation: the cytokine-hepcidin link. Journal of Clinical Investigation. 2004;113(9):1251. 367. van Stuijvenberg ME, Smuts CM, Wolmarans P, Lombard CJ, Dhansay MA. The efficacy of ferrous bisglycinate and electrolytic iron as fortificants in bread in iron-deficient school children. British journal of nutrition. 2006;95(03):532-538.

190

368. Grant FK, Suchdev PS, Flores-Ayala R, et al. Correcting for inflammation changes estimates of iron deficiency among rural Kenyan preschool children. J Nutr. 2012;142(1):105-111. 369. Aguilar R, Moraleda C, Quinto L, et al. Challenges in the diagnosis of iron deficiency in children exposed to high prevalence of infections. PloS one. 2012;7(11):e50584. 370. Schulze KJ, Christian P, Wu LS-F, et al. Micronutrient deficiencies are common in 6-to 8- year-old children of rural Nepal, with prevalence estimates modestly affected by inflammation. The Journal of nutrition. 2014;144(6):979-987. 371. Beguin Y. Soluble transferrin receptor for the evaluation of erythropoiesis and iron status. Clinica Chimica Acta. 2003;329(1):9-22. 372. Zimmermann MB, Biebinger R, Rohner F, et al. Vitamin A supplementation in children with poor vitamin A and iron status increases erythropoietin and hemoglobin concentrations without changing total body iron. The American journal of clinical nutrition. 2006;84(3):580-586. 373. Cusick SE, Tielsch JM, Ramsan M, et al. Short-term effects of vitamin A and antimalarial treatment on erythropoiesis in severely anemic Zanzibari preschool children. The American journal of clinical nutrition. 2005;82(2):406-412. 374. Roodenburg AC, West CE, Hovenierl R, Beynen AC. Supplemental vitamin A enhances the recovery from iron deficiency in rats with chronic vitamin A deficiency. British Journal of Nutrition. 1996;75(04):623-636. 375. Garc ı́a-Casal MaN, Layrisse M, Solano L, et al. Vitamin A and β-carotene can improve nonheme iron absorption from rice, wheat and corn by humans. The Journal of nutrition. 1998;128(3):646-650. 376. Layrisse M, García-Casal MN, Solano L, et al. The role of vitamin A on the inhibitors of nonheme iron absorption: preliminary results. The Journal of Nutritional Biochemistry. 1997;8(2):61-67. 377. Morley R. Nutrition and cognitive development. Nutrition. 1998;14(10):752-754. 378. Bhatnagar S, Taneja S. Zinc and cognitive development. Br J Nutr. 2001;85 Suppl 2:S139- 145. 379. Atinmo T, Mirmiran P, Oyewole OE, Belahsen R, Serra-Majem L. Breaking the poverty/malnutrition cycle in Africa and the Middle East. Nutrition reviews. 2009;67(s1):S40-S46. 380. Lockwood M, Collier P. Maternal education and the vicious cycle of high fertility and malnutrition: an analytic survey. Vol 130: World Bank Publications; 1988. 381. Beard J. Iron deficiency alters brain development and functioning. J Nutr. 2003;133(5):1468S-1472S. 382. Pollitt E, Soemantri A, Yunis F, Scrimshaw N. Cognitive effects of iron-deficiency anaemia. The Lancet. 1985;325(8421):158. 383. Eilander A, Gera T, Sachdev HS, et al. Multiple micronutrient supplementation for improving cognitive performance in children: systematic review of randomized controlled trials. Am J Clin Nutr. 2010;91(1):115-130. 384. Allen LH, Peerson JM, Olney DK. Provision of multiple rather than two or fewer micronutrients more effectively improves growth and other outcomes in micronutrient- deficient children and adults. J Nutr. 2009;139(5):1022-1030. 385. Nogueira Arcanjo FP, Santos PR, Arcanjo CP, Amancio OM, Braga JA. Use of iron-fortified rice reduces anemia in infants. J Trop Pediatr. 2012;58(6):475-480. 386. WHO. WHO Global Database on Child Growth and Malnutrition. 2012. 387. WHO. Vitamin and Mineral Nutrition Information System (VMNIS). WHO Global Database on Anaemia. 2007. 388. Greenfield PM. You can't take it with you: Why ability assessments don't cross cultures. American Psychologist. 1997;52(10):1115. 389. Van de Vijver F, Tanzer N. Bias and equivalence in cross-cultural assessment. European review of applied psychology. 1997;47(4):263-279.

191

390. Aiken LR. Assessment of intellectual functioning. Springer Science & Business Media; 2004. 391. Nga TT, Winichagoon P, Dijkhuizen MA, Khan NC, Wasantwisut E, Wieringa FT. Decreased parasite load and improved cognitive outcomes caused by deworming and consumption of multi-micronutrient fortified biscuits in rural Vietnamese schoolchildren. The American journal of tropical medicine and hygiene. 2011;85(2):333- 340. 392. DeThorne LS, Schaefer BA. A guide to child nonverbal IQ measures. Am J Speech Lang Pathol. 2004;13(4):275-290. 393. Wesley LV. Fluid intelligence and executive functions: new perspectives. Intelligence: new research : Nova Publishers; 2006:2-22. 394. Gray JR, Chabris CF, Braver TS. Neural mechanisms of general fluid intelligence. Nature neuroscience. 2003;6(3):316-322. 395. Pennington BF. Diagnosing learning disorders: A neuropsychological framework. Guilford Press; 2008. 396. Milman N. Iron in pregnancy: how do we secure an appropriate iron status in the mother and child? Annals of nutrition & metabolism. 2010;59(1):50-54. 397. Sazawal S, Black RE, Bhan MK, Jalla S, Sinha A, Bhandari N. Efficacy of zinc supplementation in reducing the incidence and prevalence of acute diarrhea--a community-based, double-blind, controlled trial. Am J Clin Nutr. 1997;66(2):413-418. 398. Kim H-Y. Statistical notes for clinical researchers: assessing normal distribution (2) using skewness and kurtosis. Restorative dentistry & endodontics. 2013;38(1):52-54. 399. Perignon M, Fiorentino M, Kuong K, et al. Stunting, Poor Iron Status and Parasite Infection Are Significant Risk Factors for Lower Cognitive Performance in Cambodian School-Aged Children. PloS one. 2014;9(11):e112605. 400. Perignon M, Fiorentino M, Kuong K, et al. Impact of Multi-Micronutrient Fortified Rice on Hemoglobin, Iron and Vitamin A Status of Cambodian Schoolchildren: a Double-Blind Cluster-Randomized Controlled Trial. Nutrients. 2016;8(1):29. 401. Organization WH. Young people's health-a challenge for society: report of a WHO Study Group on Young People and" Health for All by the Year 2000"[meeting held in Geneva from 4 to 8 June 1984]. 1986. 402. Saloojee H, Pettifor JM. Iron deficiency and impaired child development. Bmj. 2001;323(7326):1377-1378. 403. Groner JA, Holtzman NA, Charney E, Mellits ED. A randomized trial of oral iron on tests of short-term memory and attention span in young pregnant women. J Adolesc Health Care. 1986;7(1):44-48. 404. Seshadri S, Gopaldas T. Impact of iron supplementation on cognitive functions in preschool and school-aged children: the Indian experience. Am J Clin Nutr. 1989;50(3 Suppl):675-684; discussion 685-676. 405. Soemantri A, Pollitt E, Kim I. Iron deficiency anemia and educational achievement. Am J Clin Nutr. 1985;42(6):1221-1228. 406. Golub MS, Keen CL, Gershwin ME, Hendrickx AG. Developmental zinc deficiency and behavior. J Nutr. 1995;125(8 Suppl):2263S-2271S. 407. Levenson CW, Morris D. Zinc and neurogenesis: making new neurons from development to adulthood. Adv Nutr. 2011;2(2):96-100. 408. Black MM. The evidence linking zinc deficiency with children's cognitive and motor functioning. J Nutr. 2003;133(5 Suppl 1):1473S-1476S. 409. Warthon-Medina M, Moran V, Stammers A, et al. Zinc intake, status and indices of cognitive function in adults and children: a systematic review and meta-analysis. European journal of clinical nutrition. 2015. 410. Brown LA, Brockmole JR, Gow AJ, Deary IJ. Processing speed and visuospatial executive function predict visual working memory ability in older adults. Exp Aging Res. 2012;38(1):1-19.

192

411. Alpert JE, Fava M. Nutrition and depression: the role of folate. Nutrition Reviews. 1997;55(5):145-149. 412. Bottiglieri T. Folate, vitamin B12, and neuropsychiatric disorders. Nutrition reviews. 1996;54(12):382-390. 413. de Gier B, Mpabanzi L, Vereecken K, et al. Height, zinc and soil-transmitted helminth infections in schoolchildren: a study in Cuba and Cambodia. Nutrients. 2015;7(4):3000- 3010. 414. McLaren DS. Letter: Classifying nutritional disease. Am J Clin Nutr. 1974;27(2):105-106. 415. Vella V, Tomkins A, Borghesi A, Migliori GB, Ndiku J, Adriko BC. Anthropometry and childhood mortality in northwest and southwest Uganda. Am J Public Health. 1993;83(11):1616-1618. 416. Briend A, Wojtyniak B, Rowland MG. Arm circumference and other factors in children at high risk of death in rural Bangladesh. Lancet. 1987;2(8561):725-728. 417. Briend A, Maire B, Fontaine O, Garenne M. Mid-upper arm circumference and weight- for-height to identify high-risk malnourished under-five children. Maternal & child nutrition. 2012;8(1):130-133. 418. WHO. Guideline: Updates on the management of severe acute malnutrition in infants and children. Geneva: World Health Organization;2013. 419. Laillou A, Prak S, de Groot R, et al. Optimal Screening of Children with Acute Malnutrition Requires a Change in Current WHO Guidelines as MUAC and WHZ Identify Different Patient Groups. PLoS One. 2014;9(7):e101159. 420. Hall G, Chowdhury S, Bloem M. Use of mid-upper-arm circumference Z scores in nutritional assessment. Lancet. 1993;341(8858):1481. 421. Hop le T, Gross R, Sastroamidjojo S, Giay T, Schultink W. Mid-upper-arm circumference development and its validity in assessment of undernutrition. Asia Pac J Clin Nutr. 1998;7(1):65-69. 422. de Onis M, Yip R, Mei Z. The development of MUAC-for-age reference data recommended by a WHO Expert Committee. Bull World Health Organ. 1997;75(1):11-18. 423. Black R. Micronutrient deficiency: an underlying cause of morbidity and mortality. Bull World Health Organ. 2003;81(2):79-79. 424. WHO. WHO child growth standards: length/height-for-age, weight-for-age, weight-for- length, weight-forheightand body mass index-for-age : methods and development. 2006. 425. Hajian-Tilaki K. Receiver operating characteristic (ROC) curve analysis for medical diagnostic test evaluation. Caspian journal of internal medicine. 2013;4(2):627. 426. Fernández MÁL, Delchevalerie P, Van Herp M. Accuracy of MUAC in the detection of severe wasting with the new WHO growth standards. Pediatrics. 2010;126(1):e195- e201. 427. Youden WJ. Index for rating diagnostic tests. Cancer. 1950;3(1):32-35. 428. Jelliffe DB. The assessment of the nutritional status of the community (with special reference to field surveys in developing regions of the world). Monogr Ser World Health Organ. 1966;53:3-271. 429. Van Den Broeck J, Eeckels R, Vuylsteke J. Influence of nutritional status on child mortality in rural Zaire. Lancet. 1993;341(8859):1491-1495. 430. Mwangome MK, Berkley JA. The reliability of weight-for-length/height Z scores in children. Maternal & child nutrition. 2014;10(4):474-480. 431. Myatt M, Duffield A, Seal A, Pasteur F. The effect of body shape on weight-for-height and mid-upper arm circumference based case definitions of acute malnutrition in Ethiopian children. Annals of Human Biology. 2009;36(1):5-20. 432. Berkley J, Mwangi I, Griffiths K, et al. Assessment of severe malnutrition among hospitalized children in rural Kenya: comparison of weight for height and mid upper arm circumference. Jama. 2005;294(5):591-597. 433. Malina R. Normal Weight Gain in Growing Children. Healthy weight journal. 1999;13(#3).

193

434. Martins PA, Hoffman DJ, Fernandes MT, et al. Stunted children gain less lean body mass and more fat mass than their non-stunted counterparts: a prospective study. Br J Nutr. 2004;92(5):819-825. 435. Clemente AP, Santos CD, Martins VJ, Benedito-Silva AA, Albuquerque MP, Sawaya AL. Mild stunting is associated with higher body fat: study of a low-income population. J Pediatr (Rio J). 2011;87(2):138-144. 436. Delisle HF. Poverty: the double burden of malnutrition in mothers and the intergenerational impact. Ann N Y Acad Sci. 2008;1136:172-184. 437. Raiten DJ, Ashour FAS, Ross AC, et al. Inflammation and Nutritional Science for Programs/Policies and Interpretation of Research Evidence (INSPIRE). The Journal of nutrition. 2015;145(5):1039S-1108S. 438. Rosales FJ, Ritter S, Zolfaghari R, Smith J, Ross A. Effects of acute inflammation on plasma retinol, retinol-binding protein, and its mRNA in the liver and kidneys of vitamin A- sufficient rats. Journal of lipid research. 1996;37(5):962-971. 439. Beard JL, Murray-Kolb LE, Rosales FJ, Solomons NW, Angelilli ML. Interpretation of serum ferritin concentrations as indicators of total-body iron stores in survey populations: the role of biomarkers for the acute phase response. Am J Clin Nutr. 2006;84(6):1498-1505. 440. Righetti AA, Glinz D, Adiossan LG, et al. Interactions and potential implications of Plasmodium falciparum-hookworm coinfection in different age groups in south-central Côte d'Ivoire. PLoS neglected tropical diseases. 2012;6(11):e1889. 441. Brown KH. Effect of infections on plasma zinc concentration and implications for zinc status assessment in low-income countries. Am J Clin Nutr. 1998;68(2):425S-429S. 442. Mao B, Chheng K, Wannemuehler K, et al. Immunity to polio, measles and rubella in women of child-bearing age and estimated congenital rubella syndrome incidence, Cambodia, 2012. Epidemiology and infection. 2014:1-10. 443. Phiri K, Calis J, Siyasiya A, Bates I, Brabin B, van Hensbroek MB. New cut-off values for ferritin and soluble transferrin receptor for the assessment of iron deficiency in children in a high infection pressure area. Journal of clinical pathology. 2009;62(12):1103-1106. 444. Engle-Stone R, Haskell MJ, Ndjebayi AO, et al. Plasma retinol-binding protein predicts plasma retinol concentration in both infected and uninfected Cameroonian women and children. J Nutr. 2011;141(12):2233-2241. 445. Wessells KR, King JC, Brown KH. Development of a plasma zinc concentration cutoff to identify individuals with severe zinc deficiency based on results from adults undergoing experimental severe dietary zinc restriction and individuals with acrodermatitis enteropathica. J Nutr. 2014;144(8):1204-1210. 446. Epstein FH, Gabay C, Kushner I. Acute-phase proteins and other systemic responses to inflammation. New England journal of medicine. 1999;340(6):448-454. 447. Liuzzi JP, Lichten LA, Rivera S, et al. Interleukin-6 regulates the zinc transporter Zip14 in liver and contributes to the hypozincemia of the acute-phase response. Proceedings of the National Academy of Sciences of the United States of America. 2005;102(19):6843- 6848. 448. Duncan A, Talwar D, McMillan DC, Stefanowicz F, O'Reilly DSJ. Quantitative data on the magnitude of the systemic inflammatory response and its effect on micronutrient status based on plasma measurements. Am J Clin Nutr. 2012;95(1):64-71. 449. Beisel WR. Infection-induced depression of serum retinol--a component of the acute phase response or a consequence? Am J Clin Nutr. 1998;68(5):993-994. 450. Dillon D, Htet MK, Chiwile FP. The application of correction factors on serum retinol of Indonesia school children. 2010. 451. Stam T, Swaak A, Kruit W, Eggermont A. Regulation of ferritin: a specific role for interferon-alpha (IFN-α)? The acute phase response in patients treated with IFN-α-2b. European journal of clinical investigation. 2002;32(s1):79-83. 452. Eijk V. Regulation of iron metabolism in the acute-phase response: interferon γ and tumour necrosis factor α induce hypoferraemia, ferritin production and a decrease in

194

circulating transferrin receptors in cancer patients. European journal of clinical investigation. 1998;28(7):520-527. 453. Kasvosve I, Gomo ZA, Nathoo KJ, et al. Association of serum transferrin receptor concentration with markers of inflammation in Zimbabwean children. Clin Chim Acta. Vol 371. Netherlands2006:130-136. 454. Punnonen K, Irjala K, Rajamäki A. Serum transferrin receptor and its ratio to serum ferritin in the diagnosis of iron deficiency. Blood. 1997;89(3):1052-1057. 455. Thurnham DI, Mburu AS, Mwaniki DL, Wagt AD. Micronutrients in childhood and the influence of subclinical inflammation. Proceedings of the Nutrition Society. 2005;64(04):502-509. 456. Baeten JM, McClelland RS, Richardson BA, et al. Vitamin A deficiency and the acute phase response among HIV-1-infected and-uninfected women in Kenya. Journal of acquired immune deficiency syndromes (1999). 2002;31(2):243-249. 457. Stephensen CB, Gildengorin G. Serum retinol, the acute phase response, and the apparent misclassification of vitamin A status in the third National Health and Nutrition Examination Survey. Am J Clin Nutr. 2000;72(5):1170-1178. 458. Bresnahan K, Chileshe J, Tanumihardjo S. Acute phase response is associated with decreased nutrient intake and altered markers of micronutrient status in Zambian children (624.16). The FASEB Journal. 2014;28(1 Supplement):624.616.

195

196

Abstract

Undernutrition causes long-term damages on the physical and cognitive development and half of deaths among children under 5 y. Most of the current interventions are concentrated on improving nutrition among young children and mothers. However, malnutrition is also prevalent among older children and adolescents, especially in Africa and in South-East Asia. It has adverse effects on their global development because of the high requirements in energy, protein and micronutrients resulting from gro wth and brain development spurts occurring these periods of life. Our research evaluated the prevalence of malnutrition and their determinants factors among school-aged children and adolescents in Senegal and Cambodia, as well as the effectiveness of micronutrient fortified food in schools in Cambodia. A cross-sectional study was conducted on a representative sample of ~ 600 children aged from 5 to 17 years from primary state schools of Dakar area, selected through a two-stage random cluster sample (30 schools × 20 chil dren). Elevated rates of thinness (19%) and micronutrient deficiencies (iron 39%, iodine 33%, zinc 26%) were showed. The food consumption survey revealed insufficient micronutrients and energy intake, as well as contribution to total energy intake insuffic ient in proteins and excessive in lipids, which reveals poor conditions among school children in Senegal. In Cambodia, a cluster-randomized placebo-controlled double-blind trial was conducted among 2400 children aged 6-16 y to measure the impact of fortif ied rice consumed during 6 months within the school feeding program of WFP on their micronutrient status and cognitive performance. Before the intervention, poor cognitive performance was associated to stunting, iron deficiency and parasite infestation. Th e growth was positively associated to zinc status, which was negatively associated to parasite infestation. Rice fortified with vitamin A improved vitamin A status and rice with the highest iron concentration had a positive impact on cognitive performance. Data from Senegal and Cambodia were analyzed to study indicators of malnutrition among school-aged children and adolescents. According to the WHO recommendations, we suggested mid-upper-arm circumference (MUAC) cut-offs to screen for acute malnutrition am ong children above 5 years in Cambodia (15.5 cm, 16.4 cm and 18.2cm in boys and 15.4 cm, 16.6 cm and 17.9 cm in girls aged 5-7.9y, 8- 11.9 y, and 11-13.9 y). We confirmed the need to correct plasma concentrations of retinol-binding-protein and ferritin acco rding to inflammatory status, and showed the interest of a similar correction for transferrin receptor, these corrections improving th e estimation of prevalence of vitamin A and iron deficiencies in populations where inflammation is endemic. Undernutrition remains a public health issue among school-aged children and adolescents in Senegal and Cambodia. Determination of malnutrition in this age range could be improved by research on indicators taking into account intra variability populatio ns, environmental factors and specific needs of these populations. School-based interventions like the one in Cambodia have positive effects on nutrition and development of children, advocating for extension of school feeding, including in disadvantaged urba n areas. Distribution fortified food in existing school canteens is a cost-effective strategy to improve nutrition and health. More research and efforts should be shifted to the nutritional status of adolescent girls because of the risk of early childbearin g, still commo nly practiced in developing countries, and the importance of adequate nutrition during pregnancy. Investing in nutrition in this age range is an opportunity to consolidate the progress achieved in pregnant women and young children through a life-cycle approach and to optimize the development of next generations in developing countries. Key-words: malnutrition, children, school-aged children, adolescents, micronutrients, food fortification, cognition, Africa, Senegal, South-East Asia, Cambodia

Résumé La dénutrition e st responsable de dommages à long terme sur le développement physique et cognitif et de la moitié des décès chez les enfants de moins de 5 ans. La majorité des interventions est concentrée sur la nutrition des jeunes enfants et des mères. Cep endant, la malnutrition, également prévalente chez les enfants plus âgés et les adolescents, notamment en Afrique et en Asie du Sud-Est, a des conséquences sur leur développement global , en raison des besoins élevés en énergie, protéines et micronutriment s induits par les pics de croissance et de développement cérébral durant ces périodes de la vie. Nos travaux ont évalué la prévalence des malnutritions et leurs facteurs déterminants chez des enfants d’âge scolaire et des adolescents au Sénégal et au Cambo dge, ainsi que l’efficacité d’une intervention avec une alimentation enrichie en micronutriments en milieu scolaire au Cambodge. Une étude transversale fut conduite sur un échantillon représentatif de 600 enfants des écoles primaires publiques de la région de Dakar sélectionnés par un échantillonnage aléatoire en grappes. Des taux élevés de maigreur (19%) et de carences en micronutriments (fer 39%, iode 33%, zinc 26%) ont été démontrés. L’enquête de consommation alimentaire a révélé des apports en micronut riments et en énergie insuffisants, ainsi qu’une contribution à l’apport énergétique déficitaire pour les protéines et en excédent pour les lipides, révélant des conditions précaires de ces scolaires en milieu urbain au Sénégal. Au Cambodge, un e ssai con trôlé randomisé en grappes en double aveugle conduit sur 2400 enfants a permis de mesurer l’impact de riz enrichis en micronutriments consommés pendant 6 mois, dans le cadre du programme d’alimentation scolaire du PAM, sur leurs statuts en micronutriments et leurs performances cognitives. Avant l’intervention, les faibles performances cognitives étaient associées au retard de croissance, à la carence en fer et aux infections parasitaires. La croissance était positivement associée au statut en zinc, lui-même négativement associé à l’infection parasitaire. Les riz enrichis en vitamine A ont amélioré le statut en vitamine A et le riz avec la concentration en fer la plus élevée a eu un impact positif sur les performances cognitives. Les données du Sénégal et du Cambodge ont été analysées pour étudier les indicateurs de malnutrition chez ces enfants d’âge scolaire et adolescents. Comme recommandé par l’OMS, nous avons déterminé les valeurs seuils du périmètre brachial pour détecter la malnutrition aigüe chez les enfants de plus de 5 ans au Cambodge. Nous avons confirmé la nécessité de corriger les concentrations plasmatiques de retinol- binding-protein et de ferritine en fonction du statut inflammatoire et démontré l’intérêt d’une telle correction pour les récepteur s de la transferrine pour améliorer l’estimation de la prévalence des carences en vitamine A et en fer dans ces populations ou l’inflammation est endémique. La dénutrition reste un problème de santé publique chez les enfants d’âge scolaire et les adoles cents au Sénégal et Cambodge. La détermination des malnutritions dans cette tranche d’âge pourrait être améliorée par des recherches sur les indicateurs prena nt en compte la variabilité intra populations, les facteurs environnementaux et les besoins spécif iques de ces populations. Des interventions scolaires comme au Cambodge ont des effets positifs sur la nutrition et le développement, plaidant en faveur d’ une extension des programmes d’alimentation scolaire y compris dans les zones urbaines défavorisées. La distribution d’aliments fortifiés dans les cantines scolaires existantes est une stratégie peu couteuse d’amélioration de la nutrition et de la santé . La nutrition des adolescentes requiert une attention particulière en raison du risque de grossesse pré coce particulièrement élevé dans de nombreux pays du Sud. Investir dans la nutrition des enfants d’âge scolaire et des adolescents est une opportunité de cons olider le progrès accompli chez les jeunes enfants et d’optimiser le développement des générations futures dans les pays du Sud. Mots-clés : malnutrition, enfants, scolaires, adolescents, micronutriments, fortification des aliments, santé, cognition, Afrique, Sénégal, Asie du sud-est, Cambodge

197